Don't re-run job until previous prologs done

If a job was requeued while some prolog was running (e.g. waiting
  for node boot, burst buffer, PrologSlurmctld, etc.) then wait
  until that completes before considering starting the job again.