fix(ingest): Handle grouping changes by also allowing revocation

[anna-parker]

resolves #2299

preview URL: https://ingest-fix-grouping-issue.loculus.org/

Summary

This is for multi-segmented genomes only.
Handle edge case when metadata fields change in reingest and previous grouping of segments is no longer valid. Currently we would add new groups and keep old groups, potentially having ingested sequences twice with no warning for users that the metadata has been updated.

This PR adds code that will detect this edge case (grouping change) and provides code to add the new groups and revoke the old ones. As there is a potential risk of over-revocation we should only run the revoke rule under manual oversight so although the code is added here I have not added the revoke rule to the rule all . In a later PR: #2392 I will add code to send us an automated message if such a change is detected and then we can launch a job which runs this snakemake rule.

This PR also adds very basic unit tests to ensure that revise and revoke work as expected.

Screenshot

Ran locally with test data (altering metadata to break grouping and cause revision)

• Revocation is successful:
  and new sequences are added.
• Revision is successful and version is updated:
  

PR Checklist

• When run locally with engineered data (see data in test folder) this will revise and revoke sequences as desired.
• The implemented feature is covered by an appropriate test.

[theosanderson]

I think I need to get my head around the concepts a bit more in discussion then re-review. One other Q. I see [object Object] in the screenshot - do we know what's up there?

[anna-parker]

I think I need to get my head around the concepts a bit more in discussion then re-review. One other Q. I see [object Object] in the screenshot - do we know what's up there?

Ah sorry should really have made a bug for this but I did it now, I see this on main always when I revoke sequences: #2401

[theosanderson]

Thanks! Sorry for the delay!