Add subclasses of OrganismSample

[turbomam]

Summary

Determine and add subclasses of OrganismSample based on the key differentia:

• Multicellular vs single-cell — e.g., a mouse vs a pure culture vs a SAG
• Prokaryote vs eukaryote — different metadata needs (ploidy, tissue type, etc.)
• Pure culture vs individual organism — colony isolate vs whole animal/plant
• Cultured vs uncultured — following GOLD's cultured / culture_type / uncultured_type pattern

Candidate subclasses (not prescriptive)

• MulticellularOrganismSample — individual multicellular organism with heterogeneous tissues/cell types
• PureCultureSample — nominally clonal culture derived by isolation
• SingleCellSample — single cell (e.g., SAG source)

Questions to resolve

• Which differentia warrant subclasses vs slots? GOLD uses slots (cultured, culture_type, domain). Subclasses make sense only if the slot sets diverge significantly.
• Can NCBI taxonomy (via samp_taxon_id) supply some differentia implicitly (prokaryote/eukaryote, cultured/uncultured)?
• Do we need a subclass for enrichment cultures or co-cultures, which don't meet the "all cells share the same genome" criterion?

Parent issues

• Part of Model isolates/organisms as distinct from environmental Biosamples #2803
• Depends on Add OrganismSample class as subclass of Sample #2883

[turbomam]

Note: host_* slot semantics flip on OrganismSample

On Biosample, host_taxid means "the taxon of the host whose body was sampled" (e.g., human gut microbiome → host is human).

On OrganismSample, the organism is the subject. So host_taxid means "the host that the isolate was isolated from" — a pathogenic bacterium's host_taxid would be the infected host species, not the bacterium itself (which is samp_taxon_id).

This semantic difference should be considered when designing subclasses. It may warrant:

• Separate slot_usage descriptions on OrganismSample for host_taxid, host_disease_stat, specific_host, etc.
• Or explicit subclass-level documentation that clarifies the role reversal
• Or renaming via alias in the OrganismSample context (e.g., host_taxid aliased as "isolation source host taxon")

GOLD handles this cleanly because host_taxonomy_id on organism_v2 always means "the host of this organism" — there's no ambiguity because the organism is always the subject.