Correct BacDive's alpha-glucosidase substrate row without globally remapping its shared legacy ChEBI ID

Related umbrella: #286. Current source assertion defect, not a claim that the
historical 234-ID cohort is now fully chemically grounded.

The legacy bacdive_mappings.tsv physical line119 explicitly names
4-Nitrophenyl alpha-D-glucopyranoside / CAS-RN:3767-28-0 /
EC:3.2.1.20 but supplies CHEBI:54684. Current BacDive emits
EC:3.2.1.20 biolink:has_input CHEBI:54684; that source node is named
4-Nitrophenyl-alpha-D-galactoside. The current native ChEBI export contains no
exact declaration or replacement witness for CHEBI:54684.

Native CHEBI:91122 has the exact glucopyranoside IUPAC synonym and stereospecific
InChIKey IFBHRQDFSNCLOZ-ZIQFBCGOSA-N. The primary manufacturer identifies the same
molecule/structure as an alpha-glucosidase substrate:
ChEBI,
manufacturer,
EC authority.

Correct this exact row to CHEBI:91122 with full original-claim evidence and
regression coverage. Do not install a global 54684-to-91122 replacement:
three other legacy rows use 54684 for galactoside / CAS-RN:7493-95-0 /
EC:3.2.1.22 and require separate reviewed disposition. Do not infer an assay
reagent link absent from the native well metadata or rewrite CAS xrefs as
identity proof. The current method directly consumes the legacy explicit pair,
so changing unified lookup aliases alone will not fix this assertion.

Evidence: data/issue1224-quarantine-20260929.xjuS9H/issue286-glucoside-authority.1cVWUE/REVIEW.md
and complete native-authority.json SHA-256
71b06e518f68c9bc3f64aaa98a0688191787d4bb43d4007061b33ef372e68c65.
All selected scientific inputs were unchanged before/after the read-only scan.
No source edit or corrected transform has yet been performed.