My guess is that it's an ESS problem. I've posted my log on the Warp9TD forum where it disabled following a load file/display toolpath event, so hopefully that will provide clues.
I think the ESS momentarily disables the drive occasionally but not long enough to latch it permanently off. During that time, pulses are still being output so they're lost and you end up with an offset. This is my speculation, but it fits nicely with the fact that there is definitely a problem with random disable events and sometimes a loss of position.