Accumulibacter is grounded to NCBITaxon:28216 (class Betaproteobacteria) though the genus has its own id

[realmarcin]

What

kb/communities/Aalborg_East_Full_Scale_EBPR_Community.yaml:

preferred_term: Candidatus Accumulibacter phosphatis and related PAOs
term:
  id: NCBITaxon:28216
  label: Betaproteobacteria
gtdb_classification:
  gtdb_id: GTDB:c__Gammaproteobacteria

Candidatus Accumulibacter is NCBITaxon:327159. The record names a genus and
stores a class.

This is the same shape as #416, found while fixing it.

How common is it — measured, and mostly benign

I swept all 715 grounded blocks for preferred_term naming a taxon finer than
its term.id, where the grounding sits at domain, phylum, or class rank.

The raw signal is noisy and most hits are correct. A descriptive group name
over an honest high-rank id is exactly right and should not be touched:

• EET-capable Rifle aquifer Firmicutes -> NCBITaxon:1239 (Bacillota)
• Prairie Pothole methanogens -> NCBITaxon:2157 (Archaea)
• rhizosphere Actinobacteria -> NCBITaxon:201174 (Actinomycetota — a synonym,
  not a mismatch)

A naive scan also trips on host-organism names: Arabidopsis root-isolated bacteria is not a claim about Arabidopsis.

After filtering those, Accumulibacter is the one clear case left in
kb/communities, plus one arguable — Tetrasphaera-related actinobacterial PAOs
on NCBITaxon:201174, where "-related" plausibly means the phylum id is
deliberate.

So this is a short, hand-checkable list, not a sweep. Reporting the measurement
because "how many more are there" is the first question #416 raises, and the
answer turns out to be "about one".

The wrinkle

and related PAOs means the entry is a group headed by Accumulibacter, not
Accumulibacter itself. Narrowing the id to the genus would make the id precise
and the scope claim wrong, in a record where the phrase is doing real work — EBPR
communities are routinely described as Accumulibacter plus co-occurring
polyphosphate-accumulating organisms.

So the call is not purely mechanical:

1. Re-ground to NCBITaxon:327159 and reword preferred_term to drop the group,
   moving "related PAOs" into notes; or
2. keep the class id and record in notes that the group is Accumulibacter-led,
   accepting an imprecise but honest id; or
3. split into two taxonomy entries — Accumulibacter, and the related PAOs.

Option 1 or 3 gives a queryable genus. A curator should pick, since it changes
what the record asserts about membership.

Note

GTDB reclassifies Betaproteobacteria into Gammaproteobacteria, which is why the
stored c__Gammaproteobacteria is not itself an error.

Found reviewing #416.