Update to correct memory access (still problem in the result that makes the particules not moving correctly)

WIP Update fftw interface to split mpi many calls and replace direct calls to fftw by calls to our interface

Add method to ``sample'' particle positions.

I need to use it for many species of particles, so I added direct
control over the particle species name.

this is a significant change of the behavior of joint_rspace_PDF,
because I didn't realize I was doing something stupid...