Hm, AP controller really expects local time it then calculates local sidereal time based on UTC offset. I assume that you have to perform an additional sync on a known star before you can perform a goto. I agree with you not using UTC in the context of astronomy is not the very best idea.
it happens ... or not
In the mean time I got from INDI forum member Spartacus two reports. He confirms the "it happens ... or not" characteristics of the unpark. Since I did not see that any time before I assume to sort that out is a matter of working on it.
Maybe a "if (dx < 1e-3 && dy == 1e-3) " parking condition may fix that,
Yes basically this will be the solution since in case of parking this condition
if (slewcomplete && (dx > PARKTHRES || dy > PARKTHRES))
with double PARKTHRES = 0.1 was/is already in place. But before I change that, I'd like to understand why it is not zero.