Did multiple more tests, all of the combinations of:
OpenLoop:ClosedLoop
G30:G38.2
MainBoard:1HCL (probe connection)
MainBoard connected would get through my test without once sticking the probe.
1HCL connected always stuck the probe, but at random times. The most # of probes I was able to get through before the probe stuck was 781.
Also tried slowing down the probing speed, and playing with the pullback (M558 H###:####). none had noticeable affect in time before sticking, though it's so random it's hard to identify what makes a difference or what is just placebo.