Consider reducing test dataset sizes

[widdowquinn]

Summary:

In the old pyani we used Blochmannia genome example datasets (among other things), and moving to pyani-plus gives us an opportunity to update our choices to reduce test run time and target particular known (prior) issues.

For example, we could introduce Donovan Parks' virus sequences (widdowquinn/pyani#340), which would allow us to simultaneously reduce input test dataset size and confirm that our fixes for the coverage problem continue to work (and validate ANIm/other tests).

[widdowquinn]

We may want to consider dividing tests of datasets into groups (e.g. "fast" and "slow"; "viral", "bacterial", "linear genomes") etc. and running these on different schedules. For example:

• "fast" tests would run in CI on a push
• "slow" tests would run in CI on a PR merge
• "complete" tests might run monthly on CI/locally when being particularly paranoid about a change.

[peterjc]

See e.g. https://docs.pytest.org/en/7.1.x/example/markers.html#registering-markers which uses marking some tests as "slow" for example.

[widdowquinn]

This would help avoid the current duplicate push and pull-request GH Action runs we have at the moment. Running the "fast" tests only on push and the "slow"/"complete" tests on pull-request would be a useful, meaningful separation of the two activities.