Add HPC oriented scheduler options

The current distributed setup works well for single machines, even with high core counts. To enable parallelization across nodes in an HPC environment we should add other scheduling options, eg via mpi or with a shared filesystem coordinated, externally started scheduler.