Back then, it was impossible to predict the extraordinary appeal that full-frame would have in the market. (Even today I find this slightly bewildering.)
The initial appeal was as you say: to use existing Canon and Nikon lens ranges that had been designed for 135-format film. That was worth a lot to Canon and Nikon (and Sony, ex Konica Minolta, to a lesser extent), but no-one starting a digital camera system from scratch would have thought to use a legacy film format that saddled you with a load of manufacturing and cost problems in exchange for hitting some arbitrary performance point that no-one had any reason to suspect would be special.
If anything, engineers expected to be able to reduce the sensor size while retaining adequate quality, since the image sensor was more efficient than film at capturing photons. Indeed, early Kodak and Olympus Four Thirds brochures talk about this.
But since Canon had the most to gain by retaining a lens range, they made tremendous efforts to get full-frame CMOS sensors off the ground, doing crazy things like aligning multiple exposures at the lithography stage. There was a Canon ‘white paper’ floating about boasting about how expensive full-frame sensors were to make. Good advertising in retrospect!
This era (from the EOS-1Ds to the original 5D for Canon) cemented the impression in the public imagination that full-frame was something special. The cameras didn’t have a one-stop noise advantage over Nikon DX as physics predicted but more than that, because Nikon had worse sensor tech.
Then, when Nikon moved from DX to FX with the D3, they made a simultaneous leap in sensor tech. This again gave an advantage to full-frame cameras that had nothing to do with the sensor size but compounded the apparent benefit.
By now, Canon and Nikon were designing new lenses for full-frame digital, and the format had become entrenched against the odds.
The move to mirrorless would have been another opportunity to start from scratch with an optimised sensor area, but by then the full-frame size (which is no fuller than any other designed-for sensor, of course, but even the jargon works in its favour) already had a market draw that could not be denied.
Conversely, Micro Four Thirds (and APS-C and even Super 35 for cinema) now has an image problem.
And it’s a quirk of history that no-one wants anything fuller than full-frame, although those sensors exist too.
All of this works against Panasonic with the GH range, but as I suggested upthread, it also gives Panasonic a unique opportunity: they can throw tech and firmware features at the GH6 with less worry about cannibalising higher-end cameras, since the high-end cameras retain the trump card of a larger sensor for differentiation. Canon and Sony cannot so easily do that with their mirrorless cameras, since those cameras already have what many people really want – a full-frame sensor – hence the necessary ‘crippling’.
Nikon, meanwhile, has no high-end video camera range to worry about, but they have been late to recognise the value of video. I think they have belatedly understood this and will make a major effort on video in the next generation of Z cameras – potentially giving Canon and Sony a headache.
I wonder how much longer (if they ever will that is) until Nikon releases a cinema body?
Perhaps they're waiting until they really nail a video-centric mirrorless body first. Thus maybe we need to wait until at least the Nikon Z6 mk3 first.
But yes, I agree with your analysis, it is an interesting quirk of history that so called "Full Frame" has cemented its place as #1