Exact-synonym overloading is not confined to 'mesophilic': 18 keys resolve to 2-9 active classes

[realmarcin]

#444 describes mesophilic declared as oboInOwl:hasExactSynonym on 8 temperature-bin classes plus the standalone mesophilic class, and states the consumer consequence: a lexical lookup returns 9 non-equivalent classes.

That is not one term. Measured against metpo.json on current main (1,619 terms, 1,216 deprecated, 403 active), 18 strings are claimed as an rdfs:label or hasExactSynonym by more than one active class:

key	n	family	as label of	as exact synonym of
mesophilic	9	temperature	METPO:1000615	METPO:1000443-446, METPO:1000450-453
alkali tolerant	7	pH	—	METPO:1000456, 1000457, 1000458, 1000461, 1000462, 1000463, 1000464
facultative acidophile	7	pH	—	METPO:1000455, 1000459, 1000460, 1000461, 1000462, 1000463, 1003007
halotolerant	7	salinity	METPO:1000622	METPO:1000465, 1000466, 1000467, 1000469, 1000470, 1000471
alkaliphile	6	pH	—	METPO:1000457, 1000458, 1000462, 1000463, 1000464, 1003002
neutrophile	5	pH	—	METPO:1000456, 1000457, 1000461, 1000462, 1003001
acidophile	4	pH	—	METPO:1000455, 1000459, 1000460, 1003003
psychrophile	4	temperature	—	METPO:1000441, 1000442, 1000448, 1000449
acid tolerant	3	pH	—	METPO:1000455, 1000459, 1000460
extreme alkaliphile	3	pH	—	METPO:1000458, 1000463, 1000464
obligative acidophile	3	pH	—	METPO:1000455, 1000459, 1000460
psychrotolerant	3	temperature	METPO:1000618	METPO:1000442, 1000449
extreme acidophile	2	pH	—	METPO:1000455, 1000459
extreme halophile	2	salinity	—	METPO:1000468, 1000472
moderate halophile	2	salinity	—	METPO:1000467, 1000471
non-halophile	2	salinity	—	METPO:1000465, 1000469
slight halophile	2	salinity	—	METPO:1000466, 1000470
thermophile	2	temperature	—	METPO:1000447, 1000454

By family: pH 9, salinity 5, temperature 4.

Why file this separately from #444

#444 proposes moving mesophilic from confirmed exact synonym to related synonym on the 8 bin classes, keeping it only on METPO:1000615. That is the right shape of fix, but applied once it leaves 17 more, and the two largest families — pH and salinity — are not mentioned there at all. Anyone reading #444 and fixing it would reasonably believe the problem is closed.

Three sub-patterns, which may want different treatment:

1. Bin overloading — the 'mesophilic' as exact synonym on 8 temperature-bin classes overloads the term #444 shape. A broad term is an exact synonym of every numeric bin in its series (psychrophile on 4, thermophile on 2, the halophile series).
2. Label/synonym crossover — mesophilic, halotolerant, psychrotolerant are each the rdfs:label of a standalone class and an exact synonym of several bins. A consumer resolving by label gets one answer, by synonym a different set, and cannot tell which was intended.
3. Optimum/range duplication — several pH and salinity keys sit on both the optimum and the range variant of the same bin (e.g. acidophile on METPO:1000455 and METPO:1000459). Those two are not interchangeable, but no lexical lookup can separate them.

Scope, honestly

This does not currently break kg-microbe: none of the 18 strings appear verbatim in our mapping inputs, so our label-to-CURIE resolution never hits them. Filing it as a correctness matter for lexical consumers generally, not as a live breakage on our side.

Reproduce against metpo.json on main: collect lbl and every hasExactSynonym value for non-deprecated classes, lowercase and strip, and report any key claimed by more than one class.

Related: #517 (qualitative trait classes not linked to their numeric bins) is the structural cause behind sub-pattern 1 — the broad term is doing the linking work that a proper relation would do.