Having perused these message boards, it seems that at least one user has been working on his own plug-in that will do something similar to this, and as I am new, I'm open to being pointed in the direction of current efforts.
That said, I think it should be a fairly simple matter to connect the Terminology Verifier feature with the NHunspell's ability to have all correctly inflected forms a particular word, such that it could verify proper terminology usage, without all of the false errors that it currently produces.
I am in the process of creating a Hunspell dictionary/affix file pair for the Tshiluba language (DR Congo). I have successfully described most of the morphological rules for this language in the affix file. Most of what remains is to simply build a more comprehensive dictionary to which I can apply these rules.
In this language, it is the verbs that are highly inflected, but with the morphological rules I have created, the verb entries in the dictionary file consist simply of the verb stems. It is these same stems that I include as my canonical forms in my termbase.
Because the two terms (the one in the target side of my termbase & the one in the Hunspell dictionary file) are identical, connecting the two should be straightforward, not requiring any type of regular expressions or the like, in order to facilitate recognition. Indeed, Hunspell already takes care of this.
I've seen quite a lot of users suggest the need for such a thing, and I would really like to see this idea move forward. I would highly encourage SDL to look at making this an integral part of the Terminology Verifier, and not wait for a 3rd party to build a plug-in.