You do have a point to a degree with lens resolution. However, I don't know where you get 18mp from.
It's where the resolution of most lenses tends to settle around over the majority of the frame (off center) based on resolutions tests. It's a very rough figure that will vary +/-30%
Maybe you can share. And I'm sure this particular topic has been covered to high heaven in the past. If what you say is true, then if I cropped a 5Dii to a 1.6 image, and compared it to a 7D image at full size there wouldn't be any resolution difference.
This topic has been covered in a bit of a misguided way, using crop sensors as a bassis for analysis is NOT a good way of doing it. The frame of a APS-C crop sensor only covers 30% of the frame of a full frame.
The very corner resolution on a crop frame is going to be the CENTER resolution at the very center on a full frame. If we use a 50D and 5DII as examples, and the 24-105mm F4.0 IS then at 24mm f/8.0 it resolves 3400 lp/ph on full frame and 3600 lp/ph on crop at around the same position.
In other words. At 24 Megapixels it resolves 92% of the available resolution but at 39 megapixels it only resolves 72% at the center. At the corners at 39 megapixels it's using only 60% of the resolution.
The center resolution of the 24-105mm at 24mm f/8.0 is therefore around 22-24 MP (in terms of pixel density), the mid frame 17-19MP and the corners 15-18 MP.
The 24-105mm is clearly outresolved by the sensor and the gains from a 39 megapixel sensor would all be incidental. You're doubling the sensor resolution but only gaining 10% more final resolution, sure it's a difference but some would say this is wasteful.
But I find from my non-scientific tests that the 7D resolves sharper at it's full field of view as compared to a 5Dii that's been cropped to the 1.6 field of view. Also I'll point out that there are the exceptional lenses, TS-E 17mm and 24mm, which many landscape photographers use all the time, but are exceptional...meaning they are the exception to most photographers. These two lenses I have no doubt could resolve quite satisfactorily on the edges on a 36mp+ sensor. Ok, I have some doubt because I've never seen it. But I wouldn't be surprised. hows that for scientific?!
There is no doubt that there are a few gems in the Canon line-up that could handle a 39+ megapixel sensor. The issue that I'm presenting is that the majority of Canon's lenses would be wasted on such a sensor with only minor incidental gains. A sensor over 22 MP would have benefited a few niches while burdening everyone else with a doubling of processing overhead. Some would argue that this would be worth it, others would not welcome it.
Also, you have to look at the bayer pattern and see how the different color channels resolve. Green channel is always perceptibly sharper than the other two channels because it has twice the photosites. So hypothetically if you doubled the number of all the photosites, then you may have far more green than you need, but then you have red and blue photosites that are as sharp as the green was.
Right I'm not denying there are many incidental benefits to higher resolution. I'm just trying to outline the argument.
Hope that helps.