Given that people have successfully modded TrackPoints into existing MX-switch keyboards of both plate-mount and PCB-mount varieties, it's clearly not impossible. Difficult to execute in a non-sketchy manner, perhaps, but not impossible. Typically, the TrackPoint unit mounts underneath the keyboard PCB/membrane...for a mechanical keyboard, you'll need a longer stick to make a TrackPoint IV high enough, but this lengthening has been done before by modders. I'm sure there are a lot of challenges that would face someone trying to come up with a better-quality implementation of the sort of changes that these modders have done, and lots more challenges with putting the end result of that procedure into production, but there's definitely enough of a proof of concept established that it's not something unreasonable to ask or hope for.
These are complicated, ugly affairs that require no small amount of engineering effort to do right. And no small cost. Oh, sure, it only adds what - $50ish to the cost of your typical laptop. That's because of volume and longevity. TrackPoint I was introduced in 1994 - and we're only now up to TrackPoint IV. Now figure IBM was shipping a tens of thousands of laptops (or at least the requisite keyboard assembly with TrackPoint III) per year, using the same exact design for 5 years - that's several hundred thousand units they get to recover those engineering costs across. (Plus those little nubs - holy crap, the list on those is hilarious.)
Cooler Master sells a lot of keyboards, no question, but even they don't come anywhere
near the volume required to recoup the engineering costs. Nor can they get the longevity required to do it, because frankly, they're not IBM/Dell/Toshiba/etc. That's pretty much exactly why TrackPoints have never been available from anyone other than massive, massive manufacturers and why the Alps GlidePoint was such a big deal when it first came out. You need to have not only the engineering outlay, but the ability to keep a part in production essentially unchanged for years and years at a time - or a source that can do that. And no, IBM will not let you license it, nor does it solve the OTHER engineering problems.
Oh, and also there's the raw component cost.
Here's a selection of force transducers to give you some idea. (I keep calling it strain gauge, but that's actually what it is.)
This one is most accurately comparable - 0 to 1.5lbs of force, 5V supply, 1V output, pushbutton. And it's $96. I also have no idea if that's sensitive enough for the application, but I'm guessing likely not.
So yeah. $96. For a single component. That still requires the ancillary components to interpret output. And a great deal of code that has to be very,
very tight on the microcontroller (which is necessarily distinct from the keyboard so no code sharing) that either has to translate your sensor to PS/2 and/or USB HID Mouse output (which requires you do ALL momentum handling in the uC) or you have to produce a signed USB HID driver to do that host-side.
So yes, it is pretty much the definition of unreasonable to expect anyone at all to make one. And no, there is no mythical 'precedent' for it - the only scale manufacturers for the technology is Tier 1s. It's too damn expensive. Unicomp is using a license from IBM and Elonex uses a very coarse, low quality off the shelf component (what do you expect? They specialize in sub-$50 readers and sub-$100 tablets.)
