Better database transaction handling in 'feedback_callback'

[tieneupin]

Previously, the feedback_callback function made use of a single persistent SQLModel.Session instance that would be passed to every function triggered by the received messages. In the event that a database error is raised (i.e. NoResultFound, MultipleResultsFound, ...) and left uncaught, or the session is not explicitly closed in the except block, the transaction will be left open in the database in an idle_in_transaction state. This will prevent other database operations from running due to locking mechanisms (e.g. our database-wiping CLI murfey.create_db).

This PR fixes that by creating a new SQLModel.Session instance using a context manager for every incoming message. The context manager will close the connection even in the event of an unhandled exception.

[codecov]

Codecov Report

❌ Patch coverage is 11.32075% with 141 lines in your changes missing coverage. Please review.
✅ Project coverage is 55.18%. Comparing base (301bffe) to head (c54d57f).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #885   +/-   ##
=======================================
  Coverage   55.17%   55.18%           
=======================================
  Files         103      103           
  Lines       11424    11424           
  Branches     1516     1516           
=======================================
+ Hits         6303     6304    +1     
+ Misses       4784     4783    -1     
  Partials      337      337           

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[stephen-riggs]

Looks fine assuming the main bit is just indented from before. One comment about an old bit of code that can be removed rather than changed

[stephen-riggs]

This exists for the old demo setup, and should just be replaced with a log if the transport object isn't set up. Recursively calling feedback_callback seems wrong

[tieneupin]

Thanks for clarifying the bit that's related to the old demo setup. I'll remove it and merge it after.

[tieneupin]

Looks fine assuming the main bit is just indented from before. One comment about an old bit of code that can be removed rather than changed

Yes, the giant if-elif block has just been indented up until where except occurs. No rewrites beyond pre-commit formatting.