Not when tripod mounted without the chance of camera shake. Frankly a longer shutter speed and f4 should make the image sharper from my experience. That is the essence of low light photography.
No, 40D does not have micro adjustment. But that photo was taken tripod mounted, and the focus point was set to the damn middle of the body of that lens. What is in focus? In front of the lens on the table. If canon said that the lens was going nuts and fixed it, then it leaves me with only one conclusion, its the 40D body.
Did you use 40D's live view function set to 10x magnification to determine if the focus performed accurately on the location you wanted to focus? Also, did you activate mirror lockup if you were not using the live view and more importantly, did you use the remote release? The remote release is the oft-neglected item to obtain a sharp image: even if the camera is mounted on the tripod, it is no use if one cannot press the shutter release button gently (by squeezing rather than pushing down); tripod alone does not guarantee sharp image, regardless of how sturdy it is. From what I gather, the test shot was made with a slow shutter speed and at near-wide open, making photographer's interaction more demanding - see previous sentence. One more note: most lenses do not perform their best wide-open, for 24-105, that's f/4. Use even slower shutter speed with self-timer and live view or mirror lock-up at between f/5.6 and f/8.
I own a 40D also and I know how it behaves, but your 40D and my 40D might be different: I cannot cite the source at this moment since I cannot recall where exactly it is, but in a nutshell, not all the cameras and lenses are created equal, even if they bear the same packaging and same label. The factory adjustment may vary from batch to batch and let's say that your 40D may have been shipped with -2 factory microadjust while mine at +2. Add to that the lenses that are shipped with tiny variations from batch to batch. It could be that the particular combination of your 24-105 to 40D is at odds with each other. Also, a zoom lens is an embodiment of a compromise - you may find that in order for the lens to perform adequately at certain focal length, the designers of the lens may have had to compromise the performance at other focal lengths. Perhaps having the camera calibrated with the lens at the focal length you use most often or find it most useful is the way to go.
On the other hand, as microadjust implies, it is a tiny thing. In real shooting, it is negligible in most cases and something that can be easily remedied. Countless SLR and DSLR owners have shot countless great images without microadjusting or even knowing about it, even if it might have made some amount of difference in their image quality.
Take more pictures, on impulse rather than thinking too much about the techniques and details. Also, take a few hundred, at least, in that fashion. If you get one good shot out of 24 or 36 (those numbers are based on frames per roll of film), then that's excellent. More often than not, you'll find that you'll take over 100 to get the one that you truly like. That ratio for me hasn't changed since the days of film camera despite all the advancement in technology. A tightly controlled test shot, while indicative of a lens's performance, is not everything that should be to determine that particular lens.
If you think 24-105 is confounding you, just you wait until you encounter a Sigma 50mm 1.4...