I had this thought before some years ago back when I was on ADSLx. Of course, the situation was not as
bad as yours, and perhaps I was over-exaggerating. Though there were no plans for FTTx then and I thought about would happen if my CPE dies? Would I ever be able to find a good replacement with a lifespan long enough before newer technology rolled out? Though it's safe to say, modems can last a long time; 5+ years is not impossible as long as the components and capacitors are of a good quality.
I can relate to this on other topics, not just DSL. Like when you buy good computer hardware for the long run and worry about stuff becoming obsolete/discontinued within a few years, wishing you had bought the components you may have needed in foresight as opposed to hindsight without having to look in a market full of used parts (with plenty of spares and repairs) once something goes wrong. You could probably apply this to just about anything.
In the DSL market, ADSLx still makes up a reasonable chunk and I do not see it dropping off the radar any time soon. At least in the UK, LLU providers are committed to providing ADSLx services for quite some years as it was their investment. That should mean a reasonable market should be maintained, while perhaps not as profitable as FTTx, for hardware supporting ADSLx. I do not have a crystal ball (and it would certainly be a requirement here), but I think there will be decent ADSLx-supporting chipsets for a while.
I can't so much comment on the testing efforts, but both VDSL and ADSL (and variants) are still DSL technologies that use Discrete Multitone (DMT) for allocation of bits. While there are some differences in the technologies, I would imagine a poorly designed chipset would negatively impact both (in fact, I imagine more critically with VDSL). But even now if you picked a modem to use that had a similar chipset to the one you have now, I can't imagine there being any significant variance; perhaps a couple hundred Kbps maximum difference, and maybe a few more or less errors a day. Besides, it is the implementation that matters just as much as chipset (and I think B10A vs B10D is much about the integrated filter).
In closing, perhaps try and focus and worry about the now and not obsoletion of CPE. There will be alternatives, which will perform within a margin.
I am sorry for the mass of writing, though I can't promise this will be the last wall of text I write.