Since I bought Reikan FoCal ages ago, I dusted it off, installed the latest version and ran the lens through it. Above is the AFMA chart it came up with at f/1.4. Essentially it did tests at -20, -10, 0, +10, +20 (body settings, not lens settings), determined from that the optimal value was -4, and that was that. Note it looks like -4 is hardly better than 0. And the +17 I had left on body before should have been way off! I don't have any explanation as to why results seemed ok previously, unless the error with focus distance is significant. I should add, this testing was done at approximately 50x focal length (2.5m), where the software reported I was at 2.8m.
It also offered the "RGB" analysis, and interestingly it shows the channels don't meet at the same point. This likely contributes to the visible LoCA in wide open shots. As well controlled it may be, it certainly is not absent.
And finally... here's the "aperture sharpness" test. I've not run this before. There was a checkbox for something about finding optimal focus which I didn't set. I should read up what that does! As it was, I don't know if it re-focused or reused the focus from the previous testing. It is still interesting. The points are the readings, of which there was one per setting. Note this only tests the centre of the lens. I guess it could do corners too with appropriate setting, but that's beyond me at the moment.
The orange line appears to be the (smoothed) overall characteristic. Not surprising, the levels rise from wide open are are generally flat from f/2 to f/5.6+, gradually dropping off after that. More interesting is once again we have an RGB plot. This appears to be smoothed too. Interesting the blue channel seems to take a bigger dive wide open. Experience of other (Canon) lenses seem to suggest red is the one more likely to defocus relative to green and blue. I might have to repeat this with the 135L as that is a good example of such.
Overall... I don't know if this is useful for anything. I just hope the new AFMA value does help, but I've ran out of time today to test it further.