Integration tests to check for report scores

[npatki]

Descriptions

When we change around abstractions and reports, it may be the case that we accidentally change how a metric works without intending to.

To pre-empt this, should we add end-to-end integration tests for the quality and diagnostic reports? These integration tests can verify that the score is always the same for some given dataset (eg. the demo dataset).

TBD: Is there any intended randomness in the logic that could affect the score? The test may have to account for it.