GitLab

also fix template parameters

particle distribution should now be uniform.

particle file layout is read from input by particle set, and then the
`particle_solver` object gets it from there when it needs it, so:
 - user-facing code is less verbose.
 - generic operating mode is implemented only once.

We now initialize the particle file layout as required.

I get an assertion failure. I need to investigate whether it's the fact
that there's no rhs data...

broken, because I'm doing something wrong with the arrays pased to
p2p_distr_mpi::compute_distr.

something is broken though.

Code doesn't actually run though. Hopefully if I initialize the vector
field properly the problem will be fixed.

preliminary. in particular solvers cannot be a simple function, they
should handle checkpointing --- especially if Adams-Bashforth particles
are to be reimplemented with the new classes.

also add easier control of template double for if.
this way i can compile faster if i need to