BTW, I had noticed a similar thing with the 70.... it would be 99% on a test chart and then in real life it was a lot more inconsistent. When i say real life i mean deliberate focus on a high contrast scene with something as a scale in the frame to check for BF/FF issues. Weird.... in the end i put the camera to the setting from the charts and hope for the best. It is ok 'most' of the time but i still don't always trust it. I take one wide open and then at F4 or so to be sure. Seems to working wide open ok but i have trust issues with it.
PS. Can't wait to get a k20d or K-7 that remembers the lenses individually.
Keep all three if you can afford it... though I use zooms for my Pentax, with my rangefinder I have three lenses only - 28/1.9, 40/2 and 50/1.2 (pretty much the same with your setup) and though it may seem the three are similar in FL, each has their own purpose... I use the 28 for group and indoor shots, the 40 as my everyday walk around (it's the smallest of the three, similar to the 43 limited in size) and the 50 for portraits (the sensor on my rangefinder has a 1.5x crop too, btw).
I've got the DA 35mm Macro, the DA 40mm, and the FA 50/1.4.
I'm tempted to sell the 35mm or trade it for a longer macro (the Sigma 105mm, for instance), but on the other hand, the 35mm FOV is awesome on digital, and the lens is just so damn handy for a variety of shooting situations.
On the other hand, while the 40mm is so tiny and fun to use (and thankfully mine doesn't seem to have the missed focus issues that yours does), it's only 5mm off from the DA 35. Seems kinda silly to have both.