Improvements to Peer InstructionI User Interface

[sethbern]

This improves the peer instruction user interface in a few places:

• When viewing saved student justifications in async PI, it now should only 4 student responses shown at random with a show-more button, plus an AI summary when the course has an API key
• Merge the justification and LLM chat tabs into one conversation
• In synchronous PI, show separate per-vote answer counts so vote 1's total stays put
  once vote 2 begins so instructors can easily compare response counts across different votes
• Fix generated analogies ignoring the student's chosen theme and over fitting to the system prompt
• Fix justifications being saved with extra wrapper text instead of just the student's answer

[sethbern]

Hey @bnmnetp, I ran black and ruff locally and they both pass, is it possible that it is failing the tests from code outside this pr?

[bnmnetp]

Passing for me locally as well. It may be a new version of ruff is catching new things... I'll look into it.

[bnmnetp]

I'm assuming you are fully up to date with runestone main?

[sethbern]

@bnmnetp I think I was behind on a few commits from the past day or two but after syncing with main it still does not pass

[bnmnetp]

Yeah, it looks like the automated CI is running ruff 0.16 whereas our pyproject.toml file has it pinned to < 0.16

[bnmnetp]

Just pushed a change to the workflow to pin the version of ruff to 0.15.x so try to rebase and push again.

[sethbern]

Lint's fixed now, thanks! crud-tests is still failing but it looks like it's in the install step before any tests run, something with pycairo not building on the runner. Tests pass for me locally though

[bnmnetp]

OK, the crud-test install has been fixed, one more rebase and hopefully this will pass on CI as well as locally.

Thanks.

[sethbern]

Everything seems to be passing now, thanks!