LLM Autograding MVP

[SethPoulsen]

This issue is to organize the MVP for pushing a Beta version of LLM autograding to PrairieLearn.

For the absolute minimum feature set, we will need:

• a new config parameter for OpenAI API key
• an "LLM autograde" button on the manual grading page
• a server job to actually call the LLM for grading
• a feature flag to hide all functionality from most people

This will still not be usable at all by instructors, but just hid behind a feature flag for testing.

Before it will actually be usable, we will need to add some way for the LLM grader to incorporate information from Instructor-graded submissions to ensure accuracy.

[SethPoulsen]

work will be done by @Victorsss-Orz but I can't assign to him since he isn't an org member at the moment.