WIP: corrections in 1.) 'test_compute_displacements_integration' and implemented 2.) 'point_to_volume_grid_fourier'

Both changes bring some problems:
1.) The integration of a step function does not converge, even though the
algorithm compute_displacements_integration seems to be right. For a
step function deformation gradient this seems to be a problem in
principle, maybe originating from even grids in the single regions of
the step.

2.) The function works but is very slow, this is for sure a problem of
the implementation of tools.i_DFT() (the inverse Fourier trafo.) for a given
x-vector which makes the interpolation possible. This should be fixed
or we can take the 'by hand computed average' in function
point_to_volume_grid() which is much faster and leads to exactly the
same results.