Add automated tests + make targets for SSSOM producer/consumer compatibility

[turbomam]

Problem

We currently lack a Python test suite that locks expected behavior for SSSOM object_id normalization/consumption. Regressions are easy to introduce when identifier format changes.

Proposal

1. Add fixture-based tests for:
   • chromadb_semantic_mapper.write_sssom_output output identifier normalization,
   • parsing curie_map metadata from SSSOM header,
   • consumer scripts handling CURIE and IRI object IDs equivalently.
2. Add make targets:
   • make test (pytest)
   • make lint (ruff check)
   • make typecheck (optional/mypy-scoped)
3. Include a tiny synthetic SSSOM fixture with:
   • OBO IRIs,
   • non-OBO IRIs,
   • CURIEs,
   • angle-bracketed variants in legacy rows.

Acceptance criteria

• A format-change PR that breaks parsing fails locally via make test.
• Make targets are documented in make help.
• CI job runs lint + tests for touched Python files.

Related

• Needed to protect current mapping cleanup and future curies/sssom-py migration.