Once you are below the DLA of the lower-resolution sensor, you won't get any more "resolution", however the detail resolved should still look better on the higher resolution sensor, as it will be more finely delinted. Beyond the DLA, you experience diminishing returns...that means things could still look better, however as you approach the minimum aperture of the lens, the improvements of the higher resolution sensor over the lower resolution sensor will diminish (not disappear, just diminish).
All of the current APS-C sensors are diffraction-limited by f/8, however diffraction is still so low at that point that it rarely matters unless you really need to resolve something with obscenely fine detail (i.e. you want to resolve the individual barbules on each barb of a birds feathers at a distance of a few feet...you are going to need something closer to a PERFECT f/4 or wider lens along with the highest resolution sensor you can possibly get your hands on...24.1mp would be pretty excellent in that situation.)
I'm still not quite convinced. So I took a image with a 7D and a 50 mm 1.8 II, aperture 11. I scaled it down to 13.5 Mpixel and up again to 18 Mpixel. Then I compared it to the original image. Even with pixelpeeping I can't recognize that the original has better image quality.
Your test strategy is fundamentally flawed, though. The proper test would be to take a photo with the 7D @ 18mp, and with something with a lower-resolution sensor in the same form factor (such as a 450D or 500D, which have 12.2mp and 15.1mp APS-C sensors respectively.) Comparing the 7D sample to the other camera, at both 100% crop and on a scale-normal basis, should clearly demonstrate the superiority of the 7D. However, that would still be insufficient unless you use an appropriate lens.
The 50/1.8 II is a decent lens, but by no means a perfect lens even by past standards. The MTF chart for that lens
actually indicates its IQ is only in the "so-so" department for the high resolution sensors of today. You really need the majority of the MTF above the 0.8 mark to be "good", and great lenses relative to todays high density APS-C sensors need to be above the 0.9 mark to be "excellent". Anything less, and a sensor like the 7D will definitely show the weakness of the lens (something I struggle with every day...the 7D outresolves even my 100-400mm L by a fair amount...only the new Mark II primes and a couple of the zooms outresolve the 7D, but when they do, WOW do they bring out the most in that camera!) Even my 50 f/1.4 USM is incapable of resolving anywhere near enough detail for the 7D...pretty much every shot is somewhat soft, even at the optimal aperture with the best AFMA setting.
The 7D is already a very high resolution sensor, so to actually realize the benefit of having a high resolution sensor, you would need to perform that test with a better lens. Something along the lines of the new 24-70 L II, which has a phenomenal MTF
, should be capable of outresolving the 7D, which would allow you to set the aperture to the DLA of the 7D on both cameras, and compare the results. I'd bet good money the 7D would definitely produce far better results than the 450D/500D if you used a higher quality lens, at pretty much any aperture below f/22. By f/22, diffraction would probably diminish the benefits of the 7D's superior resolution by such a degree that you wouldn't see any measurable difference on a normalized basis.