Remove machinery for fetching non-core sheets and BioPortal submissions

[turbomam]

Problem

The Makefile contains substantial infrastructure for downloading secondary Google Sheet tabs, BioPortal historical submissions, and external BioPortal ontologies. This machinery is not part of the ontology build — it was used for one-time research/analysis. It adds complexity, maintenance burden, and confusion about what the build actually does.

Relates to #379 (assess non-core tabs for deletion) — once the tabs are gone, the download code has no purpose.
Relates to #378 (deprecated entities in OWL) — some of this machinery was part of the deprecation research workflow.

What to remove

1. Non-core sheet downloads (~40 lines in Makefile)

The download-all-sheets target fetches 7 sheets, but only classes.tsv and properties.tsv are used in the build. Remove download targets for:

• bactotraits.tsv
• more_synonyms.tsv
• more_classes___inconsistent.tsv
• metabolic_and_respiratory_robot.tsv
• metabolic_and_respiratory_llm.tsv

And the secondary and deprecated sections in sheets.yaml (after #379 confirms those tabs are deleted).

2. BioPortal historical submission downloads (~30 lines)

METPO_SUBMISSIONS list and download-all-bioportal-submissions target fetch 16 historical METPO OWL submissions from BioPortal. This was one-time provenance research for the ID allocation audit. Remove:

• METPO_SUBMISSIONS variable
• download-all-bioportal-submissions target
• clean-bioportal-submissions target
• external/metpo_historical/ directory references

3. External BioPortal ontology downloads (~50 lines)

NON_OLS_BIOPORTAL_ONTOLOGIES list and download-external-bioportal-ontologies target fetch 6 ontologies (D3O, EDAM, MCRO, PRIDE, AMPHIMEDON, PDO) for embedding comparison. This was one-time research for ChromaDB evaluation (#364). Remove:

• NON_OLS_BIOPORTAL_ONTOLOGIES variable
• download-external-bioportal-ontologies target
• clean-external-bioportal-ontologies target
• external/ontologies/bioportal/ directory references
• Associated SPARQL extraction rules for data/pipeline/non-ols-terms/%.tsv

4. BactoTraits/Madin MongoDB infrastructure

Assess whether the MongoDB import and reconciliation targets (import-bactotraits, import-madin, bactotraits-metpo-set-diff, etc.) are still used. If the analysis is complete, remove:

• import-bactotraits, import-bactotraits-metadata, import-madin, import-madin-metadata targets
• clean-bactotraits-db, clean-madin-db targets
• Reconciliation report targets that depend on MongoDB
• Associated Python scripts if unused elsewhere

Goal

After this cleanup, the Makefile's sheet-related code should be: download classes + properties tabs, run robot template, merge into OWL. Everything else is the ontology's own scripts and SPARQL queries.