OK, here's the update. I've got new data for 33, 181, 277 (EU), 330 (EU), and 670, which happen to be the five highest mileage cars in the study. I also added a new car, #417.
My car (670) has bounced around some, but looks like maybe it's stabilizing around 180Ah. I think that it may have taken that long for the algorithm to fully realize the damage that happened from letting the cells get too low in the fall, and if so it cost me 10 Ah (ouch!). I notice that #181 looks like it's hitting a plateau at roughly the same mileage that I did for that big flat-ish period from 16K-35K miles. Alas, it was in an accident and isn't back on the road yet, so it may be a long while before we get more data on it.
#33 is an outlier in terms of loss of capacity vs. miles, but it's about where my car is based on time. 277 (EU) is tracking #181 quite well, I'll be interested to see if it plateaus at the same point in another few thousand miles. 330 (EU) is tracking 670. The big drop in CAC around 10K miles corresponded to some very long drive days, which may also mean very low SOC, which has been known to lower the CAC. It's maybe 5K miles short of where 670 and 181 plateaued, so it will be particularly interesting after the end of the summer when it's got a few thousand more miles on it.
417 (the new addition) is a pretty low mileage car, with less than 2K on the 3.0 battery. It's got a big hole in its log file, so the ~700 miles in two day that it looks like it has is an artifact. It's doing a little worse than most of the low mileage cars, but it's also got a lot of calendar time on the battery.
Thanks everyone for the data. I'll probably hold off on doing another update for a few months, unless I get tons of logs in.