DEVICE AND METHOD FOR DETECTING A SURGE IN A COMPRESSOR AND RELOCATING A SURGE MARGIN BACKGROUND TECHNICAL FIELD Embodiments of the subject matter disclosed herein generally relate to methods and devices that relocate a surge margin after occurrence of a surge event is detected based on pattern recognition in an evolution of a discharge pressure. DISCUSSION OF THE BACKGROUND Centrifugal compressors are a class of radial-flow work-absorbing turbomachinery. In a centrifugal compressor, the pressure is increased by adding kinetic-energy/velocity to a continuous flow of fluid through rotation of a rotor or an impeller of the compressor. Centrifugal compressors are frequently used in pipeline transport of natural gas to move the gas from a production site to consumers, in oil refineries, refrigeration systems, gas turbines, etc. Centrifugal compressor's operation may be affected by the occurrence of a surge. Pressure of a flow of fluid passing through the compressor increases from a surge pressure at the input of the compressor, to a discharge pressure at the output of the compressor. A surge phenomenon occurs when the compressor cannot add enough energy to overcome the system resistance, which results in a rapid flow and discharge pressure decrease. The surge may be accompanied by high vibrations, temperature increases and rapid changes in the axial thrust. These effects may damage the compressor. Most systems including compressors are designed to withstand occasional surging. However, repeated and long lasting surges may result in catastrophic failures. The system operation during a surge event is unstable. Therefore, engineers try to operate compressors away from the compressor's stability limit, by adjusting a ratio of the pressures of fluid input into and discharged from the compressor, the fluid flow or other parameters that may be controlled. A surge margin provides a measure of how close an operating state of the compressor is to a surge state. Various parameters may be used for evaluating the surge margin. For example, a surge margin may be a ratio of a fluid flow input into the compressor which engineers consider safe (i.e., no surge is expected to occur) and a surge fluid flow at which a surge is likely to occur, all other operating conditions (e.g., a ratio of a surge pressure and a discharge pressure) except the fluid flow being the same. Figure 1 represents a diagram of a conventional system 1 including an expander 10 and a compressor 20. The conventional system 1 includes an anti-surge flow recirculation loop 30 providing a flow path from an output 32 of the compressor 20 to an input 34 of the compressor 20. Along the anti-surge flow recirculation loop 30 are located a surge detector 40 and an anti-surge valve 50. The anti-surge flow recirculation loop 30 may also include a gas cooler 60 and a flow element 70. Depending on the operating states of the anti-surge valve 50, a gas flow may be recycled from the output 32 of the compressor 20 to the input 34 of the compressor 20. When the detector detects a surge trend, the anti-surge valve 50 is operated to break the surge cycle by adjusting the flow to reverse the surge trend. Conventionally, the anti-surge control and surge detection are independent. The conventional surge detection may only trip the system. A surge shot is an event characterized by the occurrence of a surge trend. Due to potentially catastrophic effects of a surge event, it is desirable to operate the system with a sufficient surge margin to avoid occurrence of any surge event. The surge detector 40 may detect an occurrence of a surge trend by monitoring a discharge pressure (pd) at the output 32 of the compressor 20. Conventionally, a surge trend is detected when the discharge pressure decreases rapidly (i.e., based on a first order derivative relative to time of the discharge pressure). A first order derivative of the discharge pressure is calculated mechanically in the surge detector 40 in Figure 1, but it may alternatively be obtained electronically based on signal processing in an electronic surge detector described below relative to Figure 2. Figure 2 illustrates a block diagram of a conventional electronic surge detector 100. The discharge pressure (V) is input to the calculation block 110 and to the add/subtract block 120. A time constant (7) is also input to the calculation block 110. The calculation block 110 outputs a value proportional to the discharged pressure (V) obtained using a first order lag filter with time constant T. The add/subtract block 120 subtracts the discharge pressure from the value output by block 110, and outputs a value (A) that (expressed in Laplace transform nomenclature) is equal to -pdTs/(1+Ts), to the comparison block 130. The comparison block 130 sends a signal to the event counter block 140 if the value (A) received from block 120 is larger than a predetermined value (B), which is separately input to the comparison block 130. The event counter 140 keeps track of a number of signals, which represent surge shots, received from the comparison block 130 within a predetermined time interval (T3surge), whose value is entered separately to the event counter 140. If two or more surge shots occur during a period equal to the predetermined time interval (T3surge), the event counter 140 outputs an alarm signal. If three or more surge shots occur during a period equal to the predetermined time interval (T3surge) the event counter 140 outputs a trip signal, signaling imminent trip (i.e., shut down) of the system. The conventional surge detection has the disadvantage that a surge shot detection depends only on an instantaneous discharge pressure slope (i.e., the first derivative of the discharge pressure). However, a discharge pressure versus time pattern typically occurring in after the surge trend has more complex features. For example, after the discharge pressure drops abruptly in a relatively short time a minimum pressure value is reached, and then the discharge pressure increases again. Conventional recognition of this surge pattern is weak because it considers only on the first time derivative of the discharge pressure at the beginning of the surge shot. Additionally, the conventional system provides no recovery action if the anti-surge controller operates based on an erroneously configured surge line, the only response of the conventional system being tripping of the system. For example, if the margin is set too low with respect to the real surge line, the anti-surge control through the loop 30 cannot maintain a minimum safe flow through compressor and a surge trend cycle may occur at a frequency that depends also on a closure rate of the anti-surge valve 50. Another disadvantage of the conventional system 1 is that an amplification applied to the time derivative of the discharge pressure is related to the predetermined threshold used for determining the occurrence of a surge shot. Accordingly, it would be desirable to provide systems and methods that avoid the afore-described problems and drawbacks. SUMMARY According to one exemplary embodiment, a fluid transport system includes (a) a compressor configured to increase a pressure of a fluid flow passing therethrough, (b) an anti-surge flow recirculation loop configured to selectively redirect a part of the fluid flow passing through the compressor from a discharge output of the compressor to an input of the compressor, and (c) a controller connected to the anti-surge flow recirculation loop and the compressor, and configured (i) to detect a surge event based on an evolution of a discharge pressure of the compressor, a rate of the discharge pressure, and a rate of change of the rate of the discharge pressure, and (ii) to relocate a surge margin characterizing an operation of the fluid transport system, based on a surge parameter value recorded at a beginning of the surge event. According to one exemplary embodiment, a method for a fluid transport system including a compressor includes (i) detecting a beginning of a surge event based on a rate of a discharge pressure of the compressor and a rate change of the rate of the discharge pressure, (ii) after the beginning of the surge event, monitoring the pressure until the discharge pressure decreases below an expected low discharge pressure value, (iii) after the discharge pressure has decreased below the expected low discharge pressure value, detecting an end of the surge event when the rate of the discharge pressure becomes positive, and (iv) after the end of the surge event, relocating a surge margin based on a surge parameter value recorded at the beginning of the surge event. According to another embodiment, a controller has (i) an interface configured to receive values of discharge pressure from a compressor, and to output signals and alarms, (ii) a surge event detection unit connected to the interface and configured to detect a surge event in the compressor based on evolutions of the discharge pressure, a rate of the discharge pressure and a rate change of the rate, and (in) a surge margin relocation unit connected to the surge event detection unit and the interface, and configured to relocate a surge margin relative to a surge parameter value recorded at a beginning of the surge event, after the surge event detector detects a pattern of a surge event in the evolutions. BRIEF DESCRIPTION OF THE DRAWINGS The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate one or more embodiments and, together with the description, explain these embodiments. In the drawings: Figure 1 is a schematic diagram of a conventional system including a compressor and a mechanical surge detector; Figure 2 is a block diagram of a conventional electronic surge detector; Figure 3 is a graph of an evolution of the discharge pressure, when a surge trend occurs; Figure 4 is a schematic diagram of a system including a compressor according to an embodiment; Figure 5 is a flow diagram of a method for detecting a surge and relocating a surge margin according to an embodiment; Figure 6 is a flow diagram of detecting a beginning of a surge event, according to an embodiment; Figure 7 is a graph representing an evolution of the first derivative of the discharge pressure, the second derivative of the discharge pressure and a deviation of the discharge pressure from an initial value, during a surge event according to an exemplary embodiment; Figure 8 is a block diagram of an electronic circuit implementing the detecting of a beginning of a surge event, according to an exemplary embodiment; Figure 9 is a flow diagram of monitoring the discharge pressure decrease, according to an exemplary embodiment; Figure 10 is a block diagram of an electronic circuit implementing the monitoring of the decreasing discharge pressure, according to an exemplary embodiment; Figure 11 is a flow diagram of detecting an end of the surge event when the first derivative of the discharge pressure indicates that the discharge pressure increases, according to an exemplary embodiment; Figure 12 is a block diagram of an electronic circuit implementing the detecting of the end of the surge event when the first derivative of the discharge pressure indicates that the discharge pressure increases, according to an exemplary embodiment; Figure 13 is a block diagram of an electronic circuit implementing the relocation of the surge margin, according to an exemplary embodiment; Figure 14 is a block diagram of a controller according to an exemplary embodiment; and Figure 15 is a graph illustrating an effect on handling a surge event in a system including a compressor, according to an exemplary embodiment. DETAILED DESCRIPTION The following description of the exemplary embodiments refers to the accompanying drawings. The same reference numbers in different drawings identify the same or similar elements. The following detailed description does not limit the invention. Instead, the scope of the invention is defined by the appended claims. The following embodiments are discussed, for simplicity, with regard to the terminology and structure of gas systems including compressors and anti-surge flow recirculation loops. However, the embodiments to be discussed next are not limited to these systems, but may be applied to other systems that require avoiding repeated surge cycles of a turbomachme. Reference throughout the specification to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment is included in at least one embodiment of the subject matter disclosed. Thus, the appearance of the phrases "in one embodiment" or "in an embodiment" in various places throughout the specification is not necessarily referring to the same embodiment. Further, the particular features, structures or characteristics may be combined in any suitable manner in one or more embodiments. Figure 3 is a graph of an evolution of a discharge pressure, when a surge trend occurs. In the following description, a surge event designates an evolution after a surge trend is observed. A person of skill in the art understands that the opening of the anti-surge valve reverses the surge trend. A surge event may be identified based on features of a pattern representing the evolution of the discharge pressure (pd) during a surge event. At a beginning 200 of the surge event, the discharge pressure decreases rapidly. A rate of the discharge pressure increases in absolute value (the actual value being negative since the discharge pressure decreases). A rate change of the rate of the discharge pressure is also increasing in absolute value (the actual value decreases because it is negative). Thus, during a surge event, a discharge pressure drops with an amount Δpd during a time interval Δtdrop- The amount of pressure drop Δpd may be around a known percentage (e.g., 12%) of the difference between a discharge pressure and a suction pressure (i.e., the pressure at the compressor's intake) at the beginning of the surge. Given the anti-surge flow recirculation loop presence, the discharge pressure is not expected to decrease significantly below a low expected value 210. The time interval Δtdrop from when the discharge pressure starts dropping until the discharge pressure starts increasing is also usually around a known time value, for example, 2.5s from when the beginning of the surge event has been observed. If during a predetermined time interval (larger than the known value), the discharge pressure does not fall below a low discharge pressure expected value, the system may consider that no surge event requiring margin relocation has occurred. After reaching a minimum value, the discharge pressure increases, e.g., 220. When the discharge pressure increases, the rate of the discharge pressure becomes positive. Figure 4 is a schematic diagram of a system 400 including an expander 410 and a compressor 420, according to an exemplary embodiment. The system 400 includes an anti-surge flow recirculation loop 430 providing a flow path from an output 432 of the compressor 420 to an input 434 into the compressor 420. Based on an evolution of a discharge pressure at the output 432 of the compressor, a controller 440 detects a surge event. The controller 440 may verify multiple features of the discharge pressure evolution. For example, the controller 440 may detect a beginning of the surge event when a rate of the discharge pressure exceeds a predetermined value, falling fast according to a change of the rate of the discharge pressure. Then, the controller 440 may monitor the discharge pressure and the rate of the discharge pressure until the discharge pressure becomes lower than a low expected value. The controller 440 may then detect an end of the surge event when the rate becomes positive. Following a surge event, the controller 440 may output a relocation alarm signal and provide a new surge margin value for operating the compressor. When a surge trend occurs, an anti-surge valve 450 on the anti-surge flow recirculation loop 430 opens to reverse the surge trend. The anti-surge flow recirculation loop 430 may also include a gas cooler 460 and a flow measurement element 470. Figure 5 represents a flow diagram of a method 500 for surge detection and margin relocation according to another embodiment. At step S510, a beginning of a surge event is detected based on values of a rate of the discharge pressure and a rate change of the rate of the discharge pressure. At step S520, the discharge pressure decrease is monitored until the pressure falls below a low expected value. At step S530, an end of the surge event is detected when the rate of the discharge pressure indicates that the discharge pressure increases. Thus, steps S510, S520 and S530 recognize the discharge pressure evolution during a surge event. At step S540, the surge margin is relocated to avoid recurrence of the surge. In contrast with the conventional approach, where the only response to the occurrence of the surge trends was to trip the system (e.g., after three shots for the conventional electronic surge detector in Figure 2), in some of the methods and systems according to various embodiments described in this section, the surge margin is relocated, which relocation makes another occurrence of a surge event less likely (since the relocated margin is farther from the surge line than the initial surge margin). Additionally, in contrast with the conventional approach in which only a surge trend is identified (i.e., a surge shot), in some of the methods and systems according to various embodiments, a beginning of a surge event is identified using evolutions of the discharge pressure, the rate of the discharge pressure and the rate of change of the rate, then the discharge pressure is monitored until decreasing below an expected low pressure value, and a reversal of the surge trend is observed when the rate of the discharge pressure becomes positive. Thus multiple features of the patter of the evolution of the discharge pressure are recognized. Figure 6 is a flow diagram of detecting a beginning of a surge event, according to an embodiment. The steps illustrated in Figure 6 may be considered a possible implementation of step S510 of the method in Figure 5. At S552, a rate Dl and a rate change D2 are calculated. The rate Dl represents a variation of the discharge pressure in time. In one embodiment, the rate Dl may be calculated as the first derivative with respect to time of the discharge pressure. In an alternative embodiment, the rate Dl with a noise reduction first order filter, may be calculated using the Laplace transform of the discharge pressure Pd(s)9 multiplied by transfer function s/(s+l) so that Dl=Pd*s/(s+l). The rate change D2 represents a variation of the rate Dl in time. In one embodiment, the rate D2 may be calculated as the second derivative with respect to time of the discharge pressure. In an alternative embodiment, the rate change D2 may be calculated using a second order noise reduction filter. In order to determine whether a surge event is likely to occur, the rate Dl is compared with a fraction k of a maximum rate (MaxRate) at S554 (since the discharge pressure decreases, if k and Max rate are positive values, a minus sign is used). The fraction k and the maximum rate (MaxRate) have predetermined values. For example, the fraction k may be around 60%. When a surge event occurs, the discharge pressure decreases rapidly. If the rate Dl remains larger than the fraction of the maximum rate (the NO branch at S554), the discharge pressure decreases slowly and no surge event is expected. If the rate Dl is smaller than the fraction of the maximum rate (the YES branch at S554), the rate change D2 is compared with a maximum rate change (MaxRateChange) at S556. As long as the rate change D2 remains larger than (-MaxRateChange), no surge event is expected (the NO branch at S556). The second derivative is used to detect a sudden (i.e., instantaneous) fast drop of Dl which indicates beginning of a surge. If the rate change D2 exceeds the maximum rate change (the YES branch at S556), a surge event is likely to occur and the current values of the discharge pressure pd, the suction pressure ps and a surge parameter Par are stored as reference values, Pd_F, PS_F and Par_F, at S558. The surge parameter may be a ratio between the flow through compressor and the flow at which surge is known to occur at same compressor pressure ratio. Based on this definition of the surge parameter, the surge parameter is one on a surge line in a two-dimensional plot of the flow parameter versus the pressure ratio. A surge margin is the value of surge parameter below which anti-surge control opens the anti-surge valve in order to maintain the surge parameter at surge margin value. Steps S552, S554, S556, and S558 illustrated in Figure 6 accomplish detecting a beginning of a surge event. The manner in which the discharge pressure and its first and second derivatives evolve during a real surge event is illustrated in Figure 7. Plot line 601 in Figure 7 represents (in arbitrary units) the first derivative with respect to time of the discharge pressure (i.e., Dl according to one embodiment). Plot line 602 in Figure 7 represents (in arbitrary units) the second derivative with respect to time of the discharge pressure (i.e., D2 according to one embodiment). Plot line 603 in Figure 7 represents the deviation of the discharge pressure from a stored value of the initial discharge pressure (in percentages). Figure 8 is a block diagram of a circuit 700 (electronic, software, hardware or combination thereof) implementing the detecting of a beginning of a surge event according to an embodiment. Blocks 710 and 720 calculate the rate Dl and the rate change D2, respectively, based on input values of the discharge pressure pd. In blocks 730 and 740, the calculated Dl and D2 are compared with a fraction K of the maximum decrease rate (- MaxRate) and a maximum decrease rate change (-MaxRateChange), respectively. If (i) the rate Dl is less than (-K×MaxRate), and (ii) the rate change D2 is less than (- MaxRateChange), the circuit 750 sends a signal to circuit 760 triggering circuit 760 to store current values of the discharge pressure pd, suction pressure ps and surge parameter Par as reference values pd_F, Ps_F and par_F. Figure 9 is a flow diagram of monitoring the discharge pressure decrease until the discharge pressure becomes lower than a low expected pressure according to an embodiment. The steps illustrated in Figure 9 may be considered as a possible implementation of step S520 of the method in Figure 5. At S782, a timer configured to measure a time (Tsurge) since the beginning of the surge event is started. If a comparison of at S784 shows that the time (Tsurge) since the beginning of the surge event has exceeded a predetermined maximum time (MaxT) (i.e., branch YES at S784), the surge shot is unlikely to follow anymore and, therefore, the surge detection logic is reset. The predetermined maximum time (MaxT) is an estimated maximum duration of a surge event. If the comparison of at S784 shows that the time since the beginning of the surge event (Tsurge) has not exceeded the predetermined maximum time (i.e., branch NO at S784), the rate Dl is compared with a fraction/of the maximum rate (MaxRate) at S786. Steps S784 and S786 are performed until the rate Dl becomes less than (-MaxRate ×f) (i.e., branch YES at S786). The current discharge pressure pd is then compared with a low expected pressure at S788. The low expected pressure is a difference between the stored value of the discharge pressure Pd_F and an expected maximum pressure drop (MaxPFall). The expected maximum pressure drop may be a predetermined fraction g of the difference between the stored value of the discharge pressure Pd_F and the stored value of the suction pressure Ps_F (e.g., the predetermined fraction g may be 12%). If the comparison at S788 shows that the discharge pressure is not lower than the low expected value (i.e., branch NO at S788), steps S784, S786 and S788 are performed again within TSurge