[Feature] Record the MPI rank count in the output file

Is your feature request related to a problem? Please describe.

The runtime datasets in the output file cannot be converted to core time because the output file does not record how many MPI ranks produced them. object_/simulation.py stores it as mpi_size, but there is no create_dataset call in output.py that records it. As a result, two output files for the same problem will not be comparable when using different MPI rank counts.

Describe the feature you'd like.

Record the rank count alongside the timers:

# mcdc/output.py, in create_runtime_dataset
file.create_dataset("runtime/mpi_ranks", data=mcdc["mpi_size"])