Enable passing datasets to tests via command line

[SolarDrew]

Adds a pytest cli option --ds which accepts a comma-separated list of strings that can be parsed by load_dataset(). If present, these are passed to any test with the accept_cli_datasets marker in place of the fixture dataset(s) actually specified for that test in the code.

The aim here is to be able to run arbitrary datasets through the test suite to more easily find issues with new data.

[codspeed-hq]

CodSpeed Performance Report

Merging #439 will not alter performance

_{Comparing SolarDrew:tests (f277cea) with main (a34aaf4)}

Summary

✅ 9 untouched benchmarks

[Cadair]

This is neat, and will be useful. Needs documenting somewhere, though god knows where.

[Cadair]

Thanks Drew