I think you've left out an important element in this software chain - NINA. I think NINA has its own "reverse angle" option for a rotator and as I recall, they changed the polarity convention they are using sometime in the recent past. So you've probably got three cooks in the kitchen here, all trying to decide if a single boolean option should be true or false. Assuming NINA wants to control the rotator for purposes of framing, I would start there - figure out how to set the 'reverse angle' option in NINA so the rotator moves in the sky direction it wants. Start with all the 'reverse' options being false then figure out what NINA wants. Once NINA and the rotator are working well together, connect PHD2 to the rotator and see if PHD2 can correctly adjust the calibration when the rotator is moved. If not, change the 'reverse' option in PHD2. Testing should be easy:
1. Do a fresh calibration at a rotator angle of zero
2. Run a guiding sequence and make sure things look right.
3. Force NINA to change the rotator angle by a significant amount, like 45 degrees.
4. Run another guiding sequence and see what happens
5. If step 4 looks good, you're done. If not, change the 'reverse' option in PHD2 and - this is important - go back to step 1 and repeat steps 1-4.
Good luck,
Bruce