State-space matrices allocation

[SalvatoreMaraniello]

The allocation of the linear system matrices (in both lin_uvlm2d_sta.py and lin_uvlm2d_dyn.py) is inefficient as based on for loops over non-contingent array elements (e.g. influence coefficient matrix allocation). Sparsity can not be exploited here as these matrices are full. For improving performance, for loops should be rewritten in C++/Fortran and the the allocation modified so as to manipulate contingent elements.