METHOD FOR FILLING A CHAMBER WITH A VARIABLE-DENSITY PARTICLE BED The present invention relates to a method for filling a chamber with a particulate solid, allowing the density of the deposited particle bed to be adjusted. The invention relates more particularly to the loading of chemical or electrochemical, petroleum or petrochemical fixed bed reactors, with solid particles in the divided state, which may be in the form of beads, grains, cylinders, pellets, rods, or any other shape, but having relatively small dimensions. The particles may in particular be molecular sieves or catalysts in bead or pellet form or most often in the form of extrudates, of single- or multilobe type, the dimensions of which vary, according to each case, from a few tenths of millimeters to a few centimeters. This is the application to which reference is made more particularly in the rest of the present description, but the method in accordance with the invention applies to the loading of any other type of solid particles into a chamber, in which it is necessary to adjust, on demand, the density of the loaded particle bed. In the field of charging fixed bed chemical reactors with solid catalyst particles, two filling techniques are mainly singled out, namely: • the conventional technique, generally denoted by the term "sock loading", which consists in introducing, at the top of the reactor, the catalyst particles into the upper end of a generally flexible sock in order to deposit them by gravity, via the other end of the same sock, at the bottom of the reactor, then onto the fixed bed that is in the process of being formed; and • the dense loading technique that consists in introducing the catalyst particles also via the top of the reactor, then in dispersing them during their fall via a generally movable device containing suitable deflectors, located in the upper part of the chamber, so as to make them drop individually by a "rain effect" onto the loading front where they are positioned freely. The catalyst beds thus obtained are loaded in a nondense and inhomogeneous manner with the sock loading type filling technique whereas the loading is dense and homogeneous when the filling method is the rain-effect method. The devices and methods enabling rain-effect loading of fixed bed chemical reactors are described for example, in applications EP 007 854, EP 116 246, EP 769 462, EP 482 991, FR 2 766 386 and FR 2 721 900. The various rain-effect loading techniques commonly used have the advantage of using as much as possible of the space available inside the chamber, while allowing a quantity of catalyst, which may possibly range up to 20% more than with the conventional "sock loading" technique, to be introduced into a given reactor volume. The high homogeneity of fixed catalyst beds loaded in a dense manner is obtained by a highly regular distribution of the catalyst particles, promoting homogeneous flow and reaction kinetics over the whole of the catalyst bed volume, whereas the conventional technique of loading using a sock (sock loading) often leads to the formation of preferential pathways for the reaction medium that are the cause of a reduction in the yield of the reactor. The dense catalyst beds have, in addition, the advantage of not settling very much throughout their whole operating period, this settling - 3 - generally not exceeding 1 to 2% of the height of the catalyst bed, whereas it may reach 8 to 10% for a conventional loading, that is to say of sock loading type. Although the dense loading techniques currently represent a large part of the market, bringing them into general use has been held back up till now by the fact that high-density beds have a relatively higher pressure drop (AP) than non-dense fixed beds of equivalent height. As a result, in such a high-density bed, the reaction medium flow rate for a given pressure is lower than in a non-dense bed or, conversely, in order to obtain an identical reaction flow rate, it is then necessary to apply a higher pressure. In addition, one of the features of the dense loading techniques is that the density obtained in due to luck, for the most part depending only on the geometric features and the state of the surface of the catalyst particles, that is to say without control being possible by adjustment of the devices and methods used. Until now, a person skilled in the art therefore had to choose between, on the one hand, non-dense catalyst beds affected by a lack of homogeneity and a significant settling of the bed over time, and homogeneous dense catalyst beds, admittedly allowing excellent reaction yields, but having a significantly higher pressure drop (AP). FR 2 812 824 proposes a method of rain-effect loading into a liquid contained in the chamber to be loaded, making it possible to thus obtain a homogeneous and non-dense loading. Unfortunately, said liquid, which must be inert with respect to the particles to be loaded, is not always available on the loading site or the specific physicochemical properties of said particles do not allow their immersion in a liquid. There is therefore a demand for a method for filling catalytic fixed bed reactors that allows catalyst beds to be obtained that are both homogeneous and relatively low dense. Within the scope of their research aiming to improve the known methods for loading reactors, the Applicant has developed a novel filling method that makes it possible to obtain homogeneous catalyst beds having a density between that of dense beds and that of nondense beds (obtained by sock loading). This novel method is a loading method using a "solid diluent", namely a granular solid material introduced into the reactor at the same time as the catalyst and which will be removed before or during start-up of the reactor, by simply dissolving it with a suitable solvent, said solvent possibly being formed directly by the reactive load. The higher the quantity of solid diluent, removed before or during start-up of the reactor, the lower the final density of the catalyst bed. Therefore, this final density of the bed may be adjusted by modifying the proportion of solid diluent introduced with the catalyst. Consequently, one subject of the present invention is a method for filling a chamber, especially a chemical reactor, with a granular solid material at an adjustable loading density, comprising the following steps consisting in: (a) filling said chamber with a mixture of a first granular solid material (A) and a solid diluent (B) , which is a second granular solid material, having different solubility properties from the first granular solid material (A); and (b) removing all of the solid diluent (B) and leaving only the first granular solid material (A) in the chamber, by passing through the fixed bed deposited in step (a), a solvent having a zero ability to dissolve the first granular solid material (A) and a strong ability to dissolve the second granular solid material (B). The method for filling the chamber with the mixture of the two solid materials (A and B) may use the conventional technique of loading with a sock (sock loading), but preferably will use the technique known as the "rain-effect" technique, consisting in depositing the material grains over the entire surface of the bottom of the chamber, then over the surface of the catalyst bed during its formation. The rain-effect loading technique may use a device consisting of a rotating, solid or hollow, central shaft, onto which flexible strips, for example made from rubber, are placed and also distributed at various levels, which diverge from said shaft according to its rotation speed. As explained above, this loading method allows the density of the final catalyst bed to be freely adjusted within a given range of densities. The higher the volume proportion of the solid diluent (B) in the granular solid catalyst (A) + solid diluent (B) mixture, the lower the final loading density. A person skilled in the art will however easily understand that this proportion of solid diluent (B) must not exceed a certain upper limit that depends, in particular, on the shape of the catalyst particles and on the size ratio of the catalyst particles to the solid diluent. Indeed, beyond a certain solid diluent (B) portion limit, the catalyst bed will become mechanically unstable after dissolving the solid diluent, and will settle as a whole under the effect of gravity or of the flow of the reaction medium, thus creating preferential flow paths that are undesirable. The Applicant has observed during their research that the volume proportion of solid diluent (B) in the granular solid (A) + solid diluent (B) mixture should generally not exceed 20 to 30%. The terra "volume proportion" is understood to mean here the ratio of the bulk volume of the solid diluent (B) to the sum of the bulk volumes of the granular solid (A) and the solid diluent (B) . This upper limit is close to 20% when the granular solid (A) and the solid diluent (B) have approximately the same dimensions and when almost all the solid diluent particles are used to separate the catalyst particles. On the other hand, when the solid diluent has dimensions that are markedly below those of the granular solid material, and when a certain fraction of the solid diluent particles are located either in the pores or cavities of the catalyst, or in the interstices between the catalyst particles without separating these particles, the volume proportion limit may be greater than 30%. Consequently, in the method of the present invention, the degree of dilution of the first granular solid material (A) by the solid diluent (B) , that is to say the V(A)/(V(A)+V(B) ) volume ratio (with V(A) = bulk volume of the granular solid (A) and V