Fixes and documentation of a few more places where the implementation will need to be updated in the future.

Only collect time-steps for hydro and gravity for the cells that have related…

Closes #455

See merge request !605

Use a dedicated MPI type for the pcells. Also use the m-pole MPI type for the communications instead of MPI_BYTE.

Correctly move the reading pointers for the various chunks in the parallel HDF5 i/o module. Increase the hdf5 chunk size to 1MB

Use the correct definition of the mean inter-particle separation for the max RMS displacement time-step calculation.