Apparently this is such a fraught question and such an ancient topic that merely asking for a simple answer to this intractable problem appears indistinguishable from trolling. I understand if no one wants to wade into this thicket all over again.
The answer in fact can be found elsewhere on AVS, and the answer is: there is no answer.
The upshot of all the endless debates on calibrating plasmas is that plasmas cannot ever be 'calibrated' for critical work, they can only be improved from factory settings.
What all the hand wringing seems to boil down to seems to be: if you optimize for one kind of viewing condition, you will not attain optimal results for other equally likely viewing conditions.
Perfect gamma? Sure, for 1% window patterns. But your gamma response will behave differently for other size windows. Let alone full fields. Perfect primaries? Sure, but now the 75% saturations won't line up. Looking good with black backgrounds? Try again with constant-APL patterns. Perfect primaries *and* saturations? Well maybe at 100% stimulus, but you can't *also* have it at 75% stimulus. Gamut looks great at 60 Hz? Surprise! There's a massive color shift at multiples of 24Hz. In fact the white balance shifts if you even change from one gamma preset to another.
For my set, I was able to improve it in two stages.
One, the device. Get the levels right, then do some rudimentary W/B, then (crucially) lower the color control a few clicks. This was a key detail I had sort of forgotten about after years living in a pure RGB world. Once you have the set as good as it can get (people are still sunburned, but less so), then you move to step two, the outboard CMS.
Here is where you can spend days balancing the perceptual tradeoffs between hue saturation and lightness, since the primaries are just broken enough, and not all in the same way, that you can never get them *all* to be perfect. Or if you do, it was because you were doing it in one controlled scenario (see laundry list above) that does not help you with other scenarios that will naturally arise in your video stream. Anyway, you keep iterating until all the dE's collectively are as low as possible overall for all permutations of colors and conditions that you have patterns for. Obviously you are trying to a) keep it simple and avoid massive settings adjustments, b) avoid any single color that is a massively wrong outlier, and c) you are generally giving preference to skin tones, colors inside the triangle formed by your secondaries, and colors at moderate stimulus levels.
Simple, isn't it.
Anyway I got it to where things are still not great, certainly not linear, but anyway mostly *all* under de94<5 which is all I can really do.
I will say the newer version of HCFR is *so* much easier to use for all this than the old ColorHCFR...