Jobs in RESERVED state not handled correctly

In rerun_node() and update_rerun_node(), the rerun for RESERVED nodes requires cancelling the SLURM job. This is not implemented.