Improve Testing

Testing is currently very sparse. It's essentially just ensuring model loading works properly (not tested in all cases yet) and reproducing the notebooks as end2end tests to make sure a new version doesn't break indexing/searching or alter results.

This is an ongoing issue, with some objectives being:

1. Improve unit test coverage
2. Test every component of the data processing pipeline to ensure the training pipeline can be grown without breaking anything
3. Test model loading in a variety of circumstances
4. Just about anything else you can think of: it should be tested

Any contributions of even a single test would be very welcome!

Start by reviewing the existing model-loading tests and notebook-based end-to-end tests mentioned in the issue. Identify one focused gap in model loading or the data processing pipeline, add a test with a clear expected result, and run the relevant test suite to confirm it passes.

Written by the indexing model from the issue text.