Just FYI Justin, I got access to exactly your setup (5D2+24-105/4L) and took the opportunity to compare it to my 7D+17-55/2.8. Tripod, no IS, with a boring bookshelf as the subject. Manual focusing of centre with live view, remote shutter release with mirror lock up. Focal plane parallel to bookshelf. 24-105/4L @ 70mm (supposedly its worst focal length), 17-55/2.8 @ 44mm (the equivalent). iso 160 and 1s exposure for 5D2, iso 100 and 1s exposure for 7D (gives the same exposure level).
Result: Sharpness about equal in corners, 24-105/4L better in centre.
Then I stepped down the aperture 2 steps (i.e. to 8.0 and 5.6, respectively) and increased exposure time to 4s.
Result: 24-105/4L@8.0 improved corners significantly, and center somewhat, clearly out-resolving the firstname.lastname@example.org, which didn't improve as much compared to wide open.
Conclusion: The resolution of the 5D2+24-105/4L is no worse than 7D+17-55/2.8, so have no remorse for selling your EF-S lens.
(NB, I can't guarantee that my results are typical, maybe I received a good 24-105/4L copy and a poor 17-55/2.8, but what I find is in line with expectations from MTF measurements)
If you're interested I can find a way to post relevant crops of the images.