Fix that all slurm jobs were canceled if a single job failed

[daniel-wer]

Description:

• It's a bit unexpected why this was an issue, but here's the gist:
  • In a PR a couple of weeks ago I fixed that if an exception occurs inside a with executor block, shutdown doesn't wait forever for all jobs to finish which caused deadlocks (because the exception might occur before the jobs were scheduled and they won't ever finish).
  • What I didn't realize is that voxelytics relies on the fact that if a job is killed, due to a time or memory limit exception or simply fails due to another error, the cluster tools actually wait for all other jobs to finish before leaving the with block (shutdown with wait=True). This is somewhat obscure and we should think about making that more obvious and deliberate, but I restored that behavior for now.
• I added a regression test and made sure that it failed before and succeeds with this PR

Issues:

• fixes that all slurm jobs were canceled if a single job failed

Todos:

Make sure to delete unnecessary points or to check all before merging:

• Updated Changelog
• Added / Updated Tests

[github-actions]

☂️ Python Coverage

current status: ✅

Overall Coverage

Lines	Covered	Coverage	Threshold	Status
9291	7841	84%	80%	🟢

New Files

No new covered files...

Modified Files

No covered modified files...

updated for commit: d1732fa by action🐍

[valentin-pinkau]

Thanks!