#244's PR body names one free-text overlap term; the fold adds several

[realmarcin]

From the round-2 review of #244 (#254).

The PR body says one chord gains a term from free text that rejects it: CHEBI:15982 on CommunityMech–MediaIngredientMech. In fact:

• the same Cobalamine note also adds it to ProteinTraitsMech–MediaIngredientMech;
• the TAXON fold lists ELM:ELME000348 as a Fungi (NCBITaxon:4751) record on four chords, though its definition reads "from TAXON:2759 except TAXON:4751".

This is #254's class of problem. The PR body should not understate it.

Fix: describe the class and give the known instances, pointing to #254.