And, just to be clear, the message is just a warning, which will self-resolve after 30 seconds of hanging out on your screen unlike an error that will stick around until interactively dismissed.

The issue stems from how NINA detects whether an in-progress park operation has been cancelled externally to it. That's not what's happening here (the park does in fact complete) but, to the logic, it has a very brief appearance of looking as if it were cancelled. The reason why I put this logic in NINA was because a user with a RoR almost had his scope guillotined off his mount because something interrupted the park operation he needed the mount to complete before being able to (safely) close his roof. Another reason to use independent limit switches in your low-headroom RoRs, folks!

Anyway, the A-P driver was doing nothing wrong and, arguably (and based on experience with all other mount ASCOM driver) neither is NINA. The situation is really an unfortunate result of the ASCOM spec lacking a good way to detect the disposition of a mount while it is in the process of parking or finding home. It's something I'd like to bring up to Bob, Pete & Co. at ASCOMorg.