Improve SLURM integration

To efficiently perform bias adjustment on large datasets, we need the ability to do so across several nodes in an HPC environment. Our current target environment is a SLURM cluster, so we should have some form of Slurm integration for Midas.