Distribute.py script which generates a ta_list.yml file for use by download.py

[Cavtheman]

The implementation currently has two modes:

• 'Split all': splits all submissions evenly between a given number of TAs, from all student in a course
• 'By Section': splits by section, with an optional --balance flag to redistribute submissions to even the workload

Known bugs:

• Group submissions in which groups have members from different sections are duplicated, and in some cases not all members are added
• The --balance flag can cause duplicates in group assignments, since groups can be split up. The script doesn't currently consider groups at all.

Potential improvements:

• Allow redistribution to fewer/more TAs than there are sections. Currently when splitting by section, it splits into as many groups as there are sections, with no regard for the number of TAs.
• Improve the balancing algorithm such that it is not possible for a section with too many submissions to get submissions from other sections. This can happen currently, though the result is still balanced.
• Support for group assignments.

[kfl]

Thanks a lot for your PR.

However, I'll not merge this PR in its current state:

• The git history is a mess with old commits from your previous PR and a merge commit. I suggest that you either make a feature branch or rebase your fork.
• Your code does not pass the CI checks. (Currently there is really just a mypy check. I should probably add some tests as well.)
• The README should be updated with some documentation.
• I'm not super keen on merging it with known bugs (unless these are properly documented as limitations. It would also be nice to be able to detect when errors did/will occur.)

If you have difficulties fixing any of these points, feel free to either ask questions here or drop by my office.

It seems that staffeli_nt might have a longer lifespan than the two weeks I originally intended it to have. Thus, I want to maintain a minimal repo hygiene.

By the way, may I ask for which course you are using it?

[Cavtheman]

I've convinced the professor and other TAs to use it in Elements of Machine Learning. I know Sofie has also convinced the TAs in Softwareudvikling to use it

[Cavtheman]

I have made a feature branch named "distribute", but I don't think I can change the base of the pull request? I think that it should clean up the commit log.

[kfl]

It seems that you have added the --no-unzip code to this PR as well.

[kfl]

I have made a feature branch named "distribute", but I don't think I can change the base of the pull request? I think that it should clean up the commit log.

I think you have to make a new PR if you want to change branch. A PR literally means a Request to Pull from a specific branch.