not tested yet.

code runs without segfaults. I still need to test the correctness.

very simple functionality, but I gain two lines per value in the useful
file, which I consider a big deal.

This reverts commit b4bbc44b.

This reverts commit 6c803fa1.

This reverts commit f843c813.

the "simname_particles.h5" file is now only opened once, and then
samples are written the same as before.
some sanity checks of the velocity and acceleration data should still be
performed.

for linking against some particular libraries, it's necessary to add
"extra_libraries" such as '-Wl,--start-group'.
Therefore when the compile command is now constructed, "extra_libraries"
are checked to see whether they start with '-' or '/' (for static
libraries given as full file names).

Use the NOT many local size when the fftw call is split too, in order to have the correct size (usefull for corner cases only)

Use a deleter to capture the free of the unique ptr allocated with fftw - update to use only ref of the field in particle system (avoid copy or move)

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