WIP: still problems with the correction order in compute_displacements

and also in point_to_volume_grid, here I maybe should average the
nonaffine displacements instead of the absolute positions (this
prefents the problems with the PBCs).