refactor: migrate karyotype tag loading to GCS and add simulated tests

[adilraza99]

This PR implements the karyotype refactor outlined in #689.

The main change is aligning karyotype tag loading with existing data access patterns used across the codebase (e.g., site_filters). Instead of loading a bundled CSV via importlib.resources, tag SNP data is now accessed from GCS using a config-driven analysis version and the shared filesystem interface.

Changes:

• Replace importlib.resources-based loading with GCS-backed loading via self._fs
• Introduce DEFAULT_KARYOTYPE_ANALYSIS config key with optional constructor override
• Move tag data path to {version}/karyotype/{analysis}/karyotype_tag_snps.csv
• Add explicit validation for inversion names (raises ValueError for unknown inputs)
• Use contig from tag data directly instead of inversion[0:2] string slicing
• Add simulated tag SNP generation (init_karyotype_tags) and test coverage under tests/anoph/
• Remove bundled karyotype_tag_snps.csv from package resources
• Document load_inversion_tags in Ag3.rst

Note:
Simulated unit tests pass independently. Integration tests may require the tag SNP data to be available on GCS along with the corresponding config update.

Fixes #689

[adilraza99]

Hi @jonbrenas, I've made the changes. Take a look when you have a moment.

[jonbrenas]

Where does this value come from?

[adilraza99]

This is a simulated placeholder used in the test config. I’ve added a comment to make that clearer.

[jonbrenas]

The date 13/12/2023 still seems fairly random.

[adilraza99]

Good catch , the previous value was arbitrary and ended up looking like a real date, which is confusing. I have replaced it with "simtest" so it is clearly just a test value and not mistaken for an actual analysis version on GCS.

[jonbrenas]

Why are the inversions hard coded here? If we ever implement karyotypes for ever species they won't have the same inversions.

[adilraza99]

These are defined in the Ag3 simulator, so they are specific to that dataset. When karyotype support is added for other species, their simulators would define their own inversions, similar to how other simulator methods are structured.

[jonbrenas]

The actual path would probably be snp_karyotype to be consistent with the others (e.g., snp_haplotypes). Also, the name of the actual file shouldn't be hard-coded. It should probably be added to the config.

[adilraza99]

Updated the path to follow the snp_karyotype convention and made the filename configurable via the config.

[adilraza99]

Hi @jonbrenas, I’ve updated this to follow the existing patterns and addressed the review points. Let me know if this looks good.

[adilraza99]

Could you please take a look when you have a moment?
@jonbrenas

[jonbrenas]

Hi @adilraza99, I am not sure why there is an empty MANIFEST.in.

[jonbrenas]

The date 13/12/2023 still seems fairly random.

[adilraza99]

Hi @jonbrenas, thanks a lot for the review.

I've addressed the changes:

• Replaced the previous date-like value with "simtest" so it’s clearly just a test value
• Removed the empty MANIFEST.in

Please let me know if anything else needs adjustment. Thank you!