Move merged PR job directories to trash_bin_dir

[Neves-P]

This PR adds a handler function for merged PRs that then moves the job directories into a trash_bin_dir, the location of which can be set via the config.

At the moment there is no scripting or automation to delete the contents of trash_bin_dir, this functionality will follow later.

The next step is testing the current implementation on a bot instance not running in software.eessi.io:

• Merge events on PRs are detected correctly
• Directories are moved to the appropriate trash_bin_root_dir and have the correct subdirectory structure that makes trash_bin_dir.

Most likely the config files need to be adjusted and things are likely to fail as the new additions have not run within the context of a bot instance. I will do so on Hábrók in Groningen and will report back on progress here.

[trz42]

Had a look at the PR. Looks good overall. Some small suggestions here and there. I wonder if we would want to keep the cleanup_pr.sh script. It might allow us to clean up manually. We could remove it later if the whole auto-cleanup works smoothly.

[Neves-P]

As it stands now the bot will:

• Move the contents of the pr_XX directory with copytree(dirs_exist_ok = True) to trash_bin_dir. This includes the event_XXXX directory(ies) that contain the cloned PRs of the bot where the slurm.out files and other metadata also get dumped
• Reports back which directories were moved in a comment containing the directory(ies) where the files were moved to (trash_bin_dir/YYYY.MM.DD)

I think this is ready for review. I have two questions to look into as part of the review:

• Is there any case where dirs_exist_ok = True is problematic? I don't think so, as the PR numbers are unique and each PR will only be moved once, so setting this argument to True just makes it easier copying the entire directory tree without extra book-keeping, but maybe I'm wrong
• Will merging PRs that span multiple months still be copied correctly? I think so, but I'm going to test that by mimicking such a case and merging a PR to see what happens. ✅ Confirmed in Update hpc.rug.nl-2023.01-eb-4.9.0-2022b.yml Neves-Bot/software-layer#47 (comment)

[trz42]

Looked at the PR and tested it. Added a few suggestions in Neves-P#2

While the suggested changes will break some symbolic links (we can look at fixing these later), they should make the procedure efficient by only moving data around (and only fallback to copying if moving wouldn't work).

addressed

[trz42]

Looks good to me. Thanks @Neves-P !