Photozone doesn't just have different measurements for different crop factors - it makes explicit that you cannot directly compare results between sensor systems. The differences in resolution etc caused by only using the centre portion of the lens will be so negligible as to be irrelevant in the context of differences due to different AA filters, IR filters, image processing, photosite size & spacing etc etc. Even if using the same camera & different crop values, it's possible that the processing algorithms will be different.
They do note that it's not comparable across different systems, but nowhere I could find do they support your claim that pushing the center resolution harder gives negligible differences.
However, if what you say is true and the differences are completely trumped by filters and internal processing, then all tests short of a large, nearly exhaustive testing of all popular cameras, using all the different formats and output options and image processing would be all but useless without the user being able to directly match those configurations. Thus, I reject that reasoning or photozone is wasting everyone's time, especially their own.
They do not support your conclusion, nor do I.
And, of course, once you crop that way you're using fewer pixels to form the image, so of course you can't directly compare the results. So the idea that you will see any difference due to 'pushing' the lens harder remains nonsense.
It's not just an issue of cropping. How in the world does my 7D have 18Mpx with less than half the size of the area of the sensor as my 22Mpx 5d3? Shouldn't it have about 10Mpx?
The sensor density is different, and it will push a lens harder. Much harder.
But I guess I can use numbers. Let's assume green (since that has double the number of photosites on a Bayer filter) and a
50mm 1.4 lens shot at f/2 and at f/11 on a 7D and a 5D3.
The 7D has a resolution of 232 pixels/mm based on the pixels/sensor dimensions.
The 5D3 has a resolution of 160 px/mm.
For lenses, the max Resolution = wave length * f-number.
So using green, ~500nm * 2 = 0.0005mm * 2 = 0.001mm = 1000 lines per mm. That's a lot, and no sensor on the market has that kind of density. This drops as you move toward red, but doesn't really go below about 700 lines per mm.
But now let's look at max resolution at f/11
0.0005mm * 11 = 182 lines/mm.
The 5D3 mostly matches the theoretical resolution of a perfect lens (which cannot exist) at f/11, 160 vs 182 l/mm.
The 7D exceeds it, at 232 vs 182. A perfect lens is pushed harder by the 7D and cannot display as well as the 5D3.
We might argue there has to be a line between them on the sensors, in which case they would have half the lines/mm, but the points of light need to be separated, too, so I'm going to call that a wash. In any case, you can search for airy disks, circles of confusion, and the Rayleigh criteron if you want to verify how sensor density matters, and why moving to the much higher-density crop cameras will change the dynamics of the camera/lens relationship, including pushing lenses past their resolving limits.
Photozone notes this on their review of one of the best lenses made, the Canon 24mm TS-E, emphasis is mine:
The Canon delivers very impressive resolution figures at its conventional setup. The image center is already exceedingly sharp at f/3.5 and basically hitting and probably exceeding the limits of the camera sensor.
They are obviously surprised that a lens isn't out-resolved by the sensor.
As I said, you won't see a difference on movies which are generally better at telling stories when the audience isn't being distracted by unnecessary extra detail, and a 1920x1080 resolution is considerably different than 5760×3840. But at the moment, we're talking about stills here (which I make my living creating), and photozone and myself disagree with your belief that such differences are nonsense or that lenses are not being pushed to their limits.
Lenses have out-resolved film, but sensors have out-resolved most lenses. This is why there's such a push to get better lenses by all the major players. Still photographers, like myself, are seeing the problems of old lenses. Most of us don't want more megapixels, we want better lenses. Canon and Sigma have been delivering lately, too.
But I'm going to throw you a bone here. Certainly, most lenses produce better images in the center portion, so if any region of the lens is pushed, the center is invariably the best region to exploit and closest to the theoretical max. Similarly, corners are often problematic for full-frame cameras on nearly all lenses, so the full frame is extremely likely to have a worse image than a higher-density sensor pushing the good center of the lens harder, except on the very best lenses or on those with huge image circles like the TS-E series or (what I often use) medium format lenses on a smaller sensor like my 5D3.