EDIT

In Casearia at least in 1 (first) case the name that was not used in a synonym was originally in a synonym that was 1x updated but never deleted according to the AUD tables. This generally can not happen in a normal save. So either the record was deleted manually in the DB (I don't remember such an action) or something went wrong during saving the AUD table.

SELECT ev.*, tb.*
FROM TaxonBase_AUD tb INNER JOIN AuditEvent ev ON ev.revisionnumber = tb.REV
WHERE tb.name_id IN (xxx) 

AM:

I did some further research on this issue and found some issues.

Generally the problem arises when the handling of the homotypic groups is not correct as I already described below. This means that in the DB there are more names in the group then can be seen in TaxEditor taxon editor (synonymy). So there are 2 things we can do on this: (1) avoid that this happens and/or (2) remove those names and there types from the homotypic group during print publication output.

Fixing this for the output should be possible in reasonable time. We already recently had a similar issue were to many names were shown in the export (https://dev.e-taxonomy.eu/redmine/issues/10357).
This needs to be done anyway to be on the safe side.

About (1): it is still not fully clear when exactly this happens. 1 possible case is when you try to delete a name from synonymy but the name record is still used elsewhere (e.g. in a name relationship as basionym or as an earlier homonym or as nameUsedInSource, …). In this case the name is removed from synonymy but is still in the database AND is still in the homotypic group.
This is something that needs to be discussed and maybe changed. A solution is not trivial as in theory it is not forbidden to have a name in the database which belongs to the same homotypic group as other names but is for some reason (e.g. regional focus) not mentioned in the synonymy. The only thing that is definitely forbidden by taxonomic rules is a situation where 2 names belong to 2 taxa of the same classification and they share the same homotypic group (in the database) – except if 1 name is an autonym created from the other name.

In the below example e.g. a name record for the name “Samyda tomentosa Sessé & Moc.” adds additional type information to C. dentate. This name in some point of time must have been in the same homotypic group as C. dentate. Later you correctly removed it from there but for some unclear reason it was not fully removed from the database but stayed as an orphaned name including type information.

So to summarize, you probably did not do anything completely wrong but the main issue is probably a less accurate behaviour of the software.

Further issues you have to have in mind is HG handling during reuse of names: this is usually an intended situation, however the handling of the homotypic group is unclear for the user here. If you reuse a name in >1 taxon (including misapplications) for the UI perspective it is difficult to make clear what names still belong to a HG.

Also it needs to be discussed how the HG of orth. var. is to be handled should original spellings or orth. var. have there own HG in the DB or should they get the HG of the correctly spelled name? As they refer to the same type the later seems to be correct from the taxonomic point of view.

In generally it needs to be discussed if the concept of a HG which is only referenced by name records and where 1 name record can have only exact 1 HG makes sense in practice. Closer to what we do in the UI is a HG which is more a "homotypic synonym group". So a homotypic group a 1 or more names belong to in the context of a given taxon. This is mostly how we currently use HGs. Even if there are caveats in theory as (1) names can not be marked as belonging to the same HG without exising in a taxon or (2) names belonging to the same HG are not recognized as such if they are not mentioned in the same synonymy (but in 2 different one that e.g. have different geographic focus or (3) autonyms and there original names can not belong to the same HG though in reality they are based on the same types.

However, in current practice all these caveats are of no real importance and/or, if necessary, can be handled in another way. E.g. (3) autonym relations could be handled via name relationships or (2) the full group of names that might belong to a "taxonomically correct" HG can still be computed (though computation is a bit more complex). For (1) and also (2+3) we could also have a "has same type as" name relationship as we had in the Berlin Model. Though this is definitely less comfortable.

The "homotypic synonym group" may also go togehter with the idea of a syn. sec. that is attached to a HG rathre then to a single synonym. (#10403)