It's generally been my experience that when it comes to selecting cameras, at least for AF, one has to keep in mind that a fair amount of shooting is done in twilight conditions like this, where there isn't something bright, or "lit" or "a light", to focus on:
So my concern would be that I can get a good AF in these conditions (MF shooters can chuckle with confidence). I know from experience that there are some cameras that I can trust in these conditions, and some, I can't...top of my list of cameras to *not* trust in these conditions would be most if not all CDAF cameras, Rebels and the 10D. From experience. Right here. Out of the 150 or so shots that I took at or near here, in this sequence, the 10D misfocused on about a third of them. I've see it blow focus on a quarter of the shots taken in broad daylight, I've seen it misfocus on just about every damm shot I took, shooting similar scenes at longer focal lengths in even more light than this, while the G9 hit every time on the same shots. I know for a fact that the 400d, at least the one *I* had, would happily misfocus on anything that wasn't a very bright white to dark transition. "near-focus"? Maybe. But still not a good focus.
The 30D and 5D I would trust in conditions like this, having shot them both handheld in much worse and gotten decent results...and I know for sure that a D300 and D700 wouldn't have a problem, for the same reasons. A700 maybe ok, I wasn't lucky enough to have one when I could shoot it without the in-camera NR. Not the most stellar low-light focus but not awful. A200 a little soft but still reliable.
Take your gear out one afternoon and shoot in the near-shadows like this and you will find out in the space of maybe 50 to 100 shots, whether the AF is reliable or not. Unless the AF calibration is wildly-off, or nearly-dysfunctional, and assuming the F# is not too low, it's not an issue of "front-focus" or "back-focus". It's just either hitting focus or missing focus. Front/back focus is only an issue at low focus-distance/FL ratios and low F#s where the DOF is a fraction of the focus-distance, not multiples of the focus-distance. When focusing 50-100 yards away at 35mm and even F2.8, front/back focus is just not a concern, and the camera either hits focus or not.
Surprisingly the G9 focuses ok in low light, while the A650 is not quite as good as the G9 (aside from the fact that the a650s' lens is not as sharp in the corners as the G9's lens, but this isn't a problem unless you need sharp corners..to me the bigger problem is the lack of raw support)...the a650 is passably good in the shadows (it could handle the shot above) but when it gets dark, it's going to misfocus often. The G9 will focus ok as long as it can find something with reasonable contrast. The big problem that remains is that they are both noisy and have low dynamic-range even at ISO80. Compared to a decent DSLR they will not produce a very high-quality image in low light even at lowest ISO, not to mention ISO200 or 400. They simply need light for good results. But this speaks to the bigger problem in that gear can be mismatched to the application, pretty easily. 9 times out of 10 what I get out of a G9 is good enough for me, though in no way can it compete with a decent DSLR and lens in twilight conditions much less at night because of the aforementioned noise and DR problems. Would I shoot panos with it? Hell yeah. Happily. Because when I shoot panos I'm not trying to get micron-level real-world resolution, and at FOVs greater than the normal 35mm FOV it's just fine, at least for me. And it can generate just as much image-data as a D700.
I think the A650 would be great for shooting panos for most people as long as you don't need the remote-control because the A650 doesn't support the Canon remote-control API.
The big thing to remember is that a phase-detect AF system needs edge-contrast to focus (and a specific orientation, to match the PDAF sensor) while CDAF and MF base focus-quality on "overall image quality". There are scales to both of these. Some systems need a lot of edge-contrast, some don't...some measure IQ to a higher degree than others. And since I've yet to see a camera review that actually measures & tests focus-quality, the only way to know is to buy and try, or to take the word of others who have shot that camera. But in the end you still have to be happy with the results that you get. Just remember that results that look ok at 4x3" can look like crap at 8x10". It doesn't take much.