Whenever sbatch exits with an error, the error is just passed to the .err log for the invoking job, but the job is marked as COMPLETED in the sacct log. This makes it difficult to determine how much data was not processed because the jobs just weren't submitted at all. This MR wraps the subprocess.run calls for sbatch in try-except blocks that pass any stderr messages back from the subprocess to the invoking command and raises an exception instead of exiting normally. This will mark those jobs as FAILED in the sacct database instead of COMPLETED.
