Huntersdad--
I've seen some of the seemingly random "off by a good 2-3 feet" focus behavior that you posted about...and haven't been able to track it down yet either. I sent my 7D2 into Canon and while they didn't say they did any "electrical adjustments" they did say they performed focus tests and adjustments...and cleaned my sensor (I didn't notice any spots or smeary effects in my test shots...so this may have been a red herring).
It does seem to focus a bit better after getting it back...but some recent pics I took of the kids sledding displayed the same missed focus behavior (front or back focused by 2-3 feet). Interestingly, before sending my body in I had MFA'd my 50 1.4 to +5...when I got it back I had to kick it up to +15. I still need to dial in my 70-200 (which was used for the kids sledding pics) but as you say, no amount of MFA will account for such a wide margin. The odd thing is (as you mention) is that sometimes it totally misses...whereas most of my shots are on the money.
All this makes me think that issues like these are algorithmic in nature...that there is some scenario where the AF system freaks out and tracking is lost. As a BBF'er, I tend to hold one of the rear focus buttons down for a good while before taking pics. For the sledding pics, I stuck to case 1, and shot a variety of single point, both expanded modes, and a few large zone AF area modes. Here's a graph showing the "batting average" for each mode. Please note that I didn't shoot equal number of frames in each mode, which is why the bars are varying height:

IMAGE LINK: https://flic.kr/p/r4obaH
As you can see in the graphs, the keeper rate for AF Point Expansion (8pt) and Large Zone AF are a good deal lower than my keeper rates in other modes (in Case 1). I feel like I'm pretty good tracking (see Single Point AF) so maybe there is something about these two modes that isn't as robust as the other modes?
Anyways, I'm going to continue to do these analyses to see if there is a consistent pattern I can discern. The relatively low keeper rate shown in the graph above doesn't jibe with another analysis I did...which shows that my keeper rates in AF Point Expansion (8pt) and Large Zone AF was pretty good...but also note the keeper rate in AF Point Expansion (8pt) in Case 1 was also a good deal lower than other Cases.
IMAGE LINK: https://flic.kr/p/qKcnws
These graphs were generated from pics of completely different subjects (the 2nd graph was from shooting x-country skiers at close range with lots of back-to-front and then side to side motion...the first graph was primarily shooting back-to-front motion)...so--apart from the relatively low keeper rate in both graphs when using AF Point Expansion (8pt) with Case 1, there is no larger pattern I can see here that would indicate an intrinsic problem with each AF Area Mode/Case combo. (I don't put too much weight on this correlation because I didn't shoot a high volume of frames in this mode/Case combo...only 14 shots in this mode for sledding, and only 11 shots in this mode for the skiers).
HTH
PP