Any chance you could snaffle a Nikon focusing magnifier, put it on your R-D 1 eyepiece, and try a couple more rounds? I've found that my own focusing repeatability isn't always as good as I'd like!
There does seem to be a distinct statistical pattern in that lenses generally are "off" the same way (front or back) and that some are always off more than others, but there's quite a bit of scatter in your data set. There are some potential mechanical sources for this, but I'd guess that the biggest one is the little ol' wet human eyeball!
(If it makes anyone feel any better, years ago I did a bunch of similar tests on film cameras, and found that they all also showed some variability in focus accuracy -- even though ALL of them produced results that looked in focus! I suspect that part of the reason we're all discussing this more with the R-D 1 is that having digital files makes it so EASY to conduct focus-accuracy tests!)
As to what to do about the Avenon, I've been thinking about that this morning and think that you may
in fact be able to deal with it by using a thinner adapter, without much (if any) penalty in infinity-focus accuracy. I'll need to develop my idea at more length than I can do here (I'm at work and surfing RFF between job requests) but will try to write up my thoughts and post them this evening.
Basically, the gist of the idea is that changing the adapter thickness by amount X doesn't change focusing accuracy by the same amount, because it displaces the RF cam by the same amount as it displaces the lens; in effect, the RF will compensate for some of the change in adapter thickness. The amount of actual
focus shift you get will depend on the ratio between the focusing movement of a 50mm lens and that of the particular lens you're using.
I'm hoping I can actually document this by measurements with a dial indicator; will take some time to do it, though.