SPECIFICATION Title of the Invention PROCESS FOR MANUFACTURING ACROLEIN OR ACRYLIC ACID FROM GLYCERIN Background of the Invention Field of the invention The present invention relates to a process for producing acrolein and/or acrylic acid from glycerol and, more particularly, to a process for preparing acrolein by dehydration of glycerol in the presence of a novel catalyst based on salt of heteropolyacid. Description of related art Fossil resources, such as oil cuts, for the chemical industry will be exhausted in a few decades. Resources of natural and renewable origin as alternative raw materials are consequently being studied more and more. Acrolein, an important synthetic intermediate for the chemical industry is produced industrially by oxidation, in the gas phase, of propylene via the oxygen in the air in the presence of catalyst systems based on mixed oxides. Glycerol, derived from animal or vegetable oils in the production of bio diesel fuels or oleochemicals is one of the routes envisaged as a substitute for propylene, glycerol being able to produce acrolein when subjected to a catalytic dehydration reaction. Such a process makes it possible to thus respond to the concept of green chemistry within a more general context of environment protection. A method for preparing acrylic acid in one step by the oxydehydration reaction of glycerol in the presence of molecular oxygen is disclosed in WO 06/114506. The principle of the method is based on the two consecutive dehydration and oxidation reactions: CH2OH-CHOH-CH2OH → CH2=CH-CH0 + 2H2O CH2=CH-CHO + ½ O2 → CH2=CH-COOH The presence of oxygen serves to carry out an oxidation reaction, following the glycerol dehydration reaction, leading to the formation of acrylic acid from the glycerol in a single step. This method can be implemented in the gas phase or the liquid phase, with concentrated or dilute aqueous solutions of glycerol. This method for producing acrylic acid directly from glycerol is particularly advantageous because it allows synthesis in a single reactor. However, it is necessary to introduce all the molecular oxygen from the dehydration stage. This has many drawbacks, in particular the reaction in the first dehydration step risks running out of control by combustion, and furthermore, when the source of molecular oxygen is air, the reactor must be much larger because of the presence of nitrogen in the air. The use of an aqueous solution of glycerol in a two-step method has the drawback of producing, at the outlet of the first stage, a stream containing not only the acrolein produced and the by-products, but also a large quantity of water, originating partly from the glycerol solution, and partly from the water produced by the dehydration reaction. Use of aqueous solutions of glycerol, however, is preferable from economic reasons. This stream is sent to the second reactor, where the acrolein is oxidized to acrylic acid in the presence of a catalyst. The conventional catalysts for this oxidation reaction are generally solids containing at least one element selected from Mo, V, W, Re, Cr, Mn, Fe, Co, Ni, Cu, Zn, Sn, Te, Sb, Bi, Pt, Pd, Ru, Rh, present in metal form or oxide, nitrate, carbonate, sulphate or phosphate form. Certain elements, such as molybdenum, tellurium or rhenium, are volatile, particularly in the presence of water. This means that the second stage catalyst looses its efficiency and its mechanical strength rapidly in the presence of the stream of water, making the maintenance of the method difficult. Moreover, the acrylic acid, produced in a dilute aqueous solution, requires separation and concentration steps that are generally complicated and fairly costly. Numerous catalyst systems have already been the subject of studies for the dehydration reaction of glycerol to acrolein. A process is known from French Patent FR 695 931 for preparing acrolein from glycerol according to which acid salts having at least three acid functional groups or mixtures of these salts are used as catalysts. The preparation of these catalysts consists in impregnating, for example with iron phosphate, pumice that has been reduced to pea-sized fragments. According to the teaching of the patent, the yield obtained with this type of catalyst is greater than 80%. In US Patent US 2,558,520, the dehydration reaction is carried out in gas/liquid phase in the presence of diatomaceous earths impregnated with phosphoric acid salts, in suspension in an aromatic solvent. A degree of conversion of glycerol to acrolein of 72.3% is obtained under these conditions. US Patent US 5,387,720 discloses a process for producing acrolein by dehydration of glycerol in liquid phase or.in gas phase at a temperature ranging up to 340°C, over acidic solid catalysts that are defined by their Hammett acidity. The catalysts must have a Hammett acidity below +2 and preferably below -3. These catalysts correspond, for example, to natural or synthetic siliceous materials, such as mordenite, montmorillonite and acidic zeolites; supports, such as oxides or siliceous materials, for example alumina (A12O3), titanium oxide (TiOi), covered by monobasic, dibasic or tribasic inorganic acids; oxides or mixed oxides such as gamma-alumina, ZnO/A12O3 mixed oxide, or heteropolyacids. The use of these catalysts would make it possible to solve the problem of formation of secondary products generated with the iron phosphate type catalysts described in the aforementioned document FR 695,931. According to International Application WO2006/087084, the strongly acidic solid catalysts whose Hammett acidity H0 is between -9 and -18 have a strong catalytic activity for the dehydration reaction of glycerol to acrolein and are deactivated less quickly. However, the catalysts recommended m the prior art for producing acrolein from glycerol generally lead to the formation of by-products such as hydroxypropanone, propanaldehyde, acetaldehyde, acetone, addition products of acrolein to glycerol, polycondensation products of glycerol, cyclic glycerol ethers, but also phenol and polyaromatic compounds which originate from the formation of coke on the catalyst and therefore from its deactivation. The presence of the by-products in acrolein, especially propanaldehyde, poses numerous problems for the separation of acrolein and requires separation and purification steps which lead to high costs for the recovery of the purified acrolein. Furthermore, when acrolein is used for producing acrylic acid, the propanaldehyde present may be oxidized to propionic acid that is difficult to separate from acrylic acid, especially by distillation. These impurities that are present greatly reduce the field of application of the acrolein produced by dehydration of glycerol. The Applicant Company has therefore sought to improve the production of acrolein from glycerol, by using more selective catalysts that make it possible to obtain high yields of acrolein and that have an activity over long durations. In the field of catalysts, French Patent FR 2 657 792 discloses a catalyst of general formula FePxMeyOz, in which: Me represents at least one of the following elements: Li, Na, K, Rb, Cs, Mg, Ca, Sr and Ba; X has a value of 0.2 to 3.0; y has a value of 0.1 to 2.0; and z is the amoumt of oxygen bonded to the other elements and that corresponds to their oxidation state, this catalyst being combined with a support, characterized by the fact that said support is a fully impregnable macroporous support having a specific surface area less than or equal to 1 m2/g, a pore volume between 0.2 and 1 cm2/g and an average pore diameter greater than or equal to 1 micron, and that the active material is deposited on the surfece of all the pores of said support, said catalyst being in the form of support grains impregnated with active material, which have a size between 0.5 and 10 mm. French Patent FR 2 498 475 teaches to use a catalyst support to which a phosphate has been added by physical mixing with the catalyst that contains a phosphate, thus making it possible to partly solve the problem of extraction of phosphate during the use of the catalyst in the preparation of methacrylic acid from isobutyric acid and oxygen. WO2007/058221 discloses a process for producing acrolein by dehydration reaction of glycerin in gas-phase in the presence of heteropolyacid used as a solid acid catalyst. The heteropolyacid is those of Group 6 element such as tungstosilicic acid, tungstophosphoric acid and phosphomolybdic acid. These heteropolyacids are supported on bi-elemental pore silica carrier and produce acrolein at a yield of 86%. This dehydration reaction of glycerin, however, is effected without oxidation gas but using nitrogen stream as carrier gas, so that deposition of carbon increase seriously and hence there is a problem of deterioration in time of stability, activity and selectivity of the catalysis. Tsukida et al. "Production of acrolein from glycerol over silica-supported heteropoly acid" CATALYSIS COMMUNICATIONS, vol. 8, no. 9, 21 July 2007, pp 1349-1353 , and Chai et al., "Sustainable production of acrolein: gas phase dehydration of glycerol over 12-tungustophosphotic acid supported on ZrO2 and S1O2", GREEN CHEMISTRY, vol.10, 2008, pp.1087-1093, and Chai et al., "Sustainable production of acrolein: preparation and characterization of zirconia-supported 12-tungustophosphotic acid catalyst for gas phase dehydration of glycerol", APPLIED CATALYST A: GENERAL, vol. 353, 2009, pp.213-222 disclose that silica or zirconia-supported heteropoly acid is effective as a catalyst for dehydration of glycerol. In WO2006/087083, oxygen is introduced to prevent degradation of the catalyst in the gas-phase reaction of glycerin. In WO2006/087084, the catalyst possessing the acid strength of HO of -9 to -18 is used. A variety of solid acid catalysts such as phosphoric acid/zirconia, Nafion/silica, sulfuric acid/zirconia, tungsten/zirconia are used in Examples and the highest yield of acrolein of 74% was obtained when tungstated zirconia catalyst was used. However, there is no catalyst usable in the industrial scale at higher performance. Inventors of this application have made a variety of studies to solve the problems and found that acrolein and acrylic acid can be produced at high yield by using salt of heteropolyacid, in which protons in a heteropolyacid are exchanged at least partially with at least one cation selected from elements belonging to Group 1 to Group 16 of the Periodic Table of Elements, and completed this invention. An object of this invention is to provide a process for producing acrolein and acrylic acid from glycerin that is a material not derived from petroleum, at a high yield. Summary of the Invention This invention is characterized by following features (1) to (16) taken separately or in combination: (1) Process for preparing acrolein by dehydration of glycerin, carried out in the presence of a catalyst comprising as a main component, at least one compound in which protons in a heteropolyacid are exchanged at least partially with at least one cation selected from elements belonging to Group 1 to Group 16 of the Periodic Table of Elements. (2) The salt of heteropolyacid is represented by the general formula (1): HaAb [X1YcZdOe] • nH2O (I) in which H is hydrogen, A is at least one cation selected from elements belonging to Group 1 to Group 16 of the Periodic Table of Elements except hydrogen, X is P or Si, Y is at least one element selected from the group comprising W, Mo, Ti, Zr. V. Nb. Ta, Cr, Mn. Fe. Co. Ni, Cu. Zn. Ga, In. Tl, Sn and Pb, Z is at least one element selected from the group comprising W, Mo. Ti. Zr, V. Nb. Ta. Cr. Mn. Fe, Co, Ni. Cu. Zn. Ga, In. Tl. Sn and Pb, and a, b, c and d being in following ranges: 0