LLM Error & Failure Reporting

[sjd210]

This PR covers a few different related changes:

• If there has been an error within the OpenAI client (e.g. out-of-date key, timeout, running out of credits), currently we are returning a zero mark answer to the user and not displaying that there has been an error. This is potentially very confusing, as the error is (almost certainly) nothing to do with the user's answer itself, but they are lead to believe it has specifically been marked wrong. To be consistent with the symbolic validator when such an error occurs, we now throw a ValidatorUnavailableException and consequently show a "Your answer could not be checked. Please try again." popup in the frontend.
• Null checking for maxMarks, with appropriate feedback (adapted from Add null checking to LLM Free Text Questions #691)
• Some wording changes for comments and error clarity, such as stating the current maximum answer length when exceeded
• More tests for various erroneous behaviours

To note - I have left default zero mark results in for cases where the OpenAI client has successfully responded but not in a format we're expecting (either not json, or using more than one message). These cases are not independent from the user's answer since they may have been caused by prompt injection or other unusual input, and we want these types of answers to still receive no marks. In this case I believe it is more helpful than an error, although I'm open to suggestions.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 35.88%. Comparing base (ea71def) to head (87f49f7).
Report is 3 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #692      +/-   ##
==========================================
+ Coverage   35.76%   35.88%   +0.12%     
==========================================
  Files         529      529              
  Lines       23478    23484       +6     
  Branches     2849     2849              
==========================================
+ Hits         8396     8427      +31     
+ Misses      14222    14196      -26     
- Partials      860      861       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.