Your latency should not be an issue. The servo loop is set at 1 mS so that's the absolute best PP could even recognize that the accessory line has changed state. My measurements show it's more like 15 mS and that's why you pretty much have to go dead slow to get an accurate reading.
In general the probing routines use the feedrate setting in PP as rough probe rate, and then a fine probe rate that is 1/20 of the that rate. Probing feedrate is limited to 60 ipm max so that means the fine rate will b 3 ipm. My testing shows you get a 0.0007" error at this rate. If I drop down to 30 ipm on the feedrate the error is 0.0003". Since this is much better than the slop in the machine who cares!
I should also point out that some of the probing routines such as pocket or bore center these errors cancel out and you're only left with the machine slop to contend with.
Yeah, that would be a problem. Since I like to tinker I would add a relay to switch between homing or ETS/probing functions. It a pretty simply fix in the homing code to bang the relay.