Refactor of particle initialisation

[Carpenteri0]

Mostly backend aesthetic changes. Main change is rejection functions have their own module, with type definitions, a few helper subroutines, and all existing rej funcs that were previously in ~3 other modules. Other smaller changes :

• Removed num_re param. Now rep and epf both use n_particles_total to set weights
• Reduced code duplication in 3 initialiser subroutines - evaluation of rej f now has its own subroutine in the rej f module
• Initialiser function is checked against particle type rather than coupling scheme - more flexibility / reuseability in initialiser subroutines between coupling schemes.
• Removed hard coded current limits in current_pdf - now finds these limits once per run before initialisation
• Also added Timo's "experimental" method of initialising particles in mod_import_experimental_dist (which is now moved to particles/initialisers/)

[N-Schwarz]

Hi @Carpenteri0 , you should be able to add reviewers to the pull request once it's ready.