Resolution again... Pixel peeping should be allowed with a license only.
No, not pixel peeping at all, but resolution; yes.
I can see the difference between the 40mm pancake and the 24-105 in a moderately sized print - sometimes. It depends on so many factors. In the centre of the frame the zoom can almost be a match, but not all pictures are about the centre of the frame and that is where the weakness comes in.
I've stated before, at Building Panoramics many of our pictures have been taken on various copies of the 24-105, and as we stitch portrait format frames with a good overlap, always close to or at f8, there is no real discernible difference between this lens and others such as the 35L, 135L, 40mm pancake, 50mm 1.4 even on large 3 metre wide prints.
However if I take a single frame landscape shot with the 24-105 at 24mm, where the detail is small and far away, I am disappointed with the result when compared with a 24mm prime or the new 24-70 L IS. I don't need pixel peeping to see the difference. I would never buy one more expensive lens compared with another if I could only see the difference at 100% on screen.
The 'tests' you have done play to the strength of the 24-105. You have shot in its best focal length ranges at a stopped down aperture of subjects that are quite close to you and relatively easy to resolve, and you have compared it against ultra fast aperture primes which are not the best lenses for stopped down photography ( in the corners ) anyway. Then of course there is the huge compromise with the image size.
I think the 24-105 is a pretty good general purpose lens, but I know where its weaknesses are and try to avoid them. I thought your picture of the temple tower in the misty dusk that you posted as taken with the 24-105 was very good, and I didn't have to pixel peep at that