For engineers who are not that familiar with pupil aberration, here is a quick summary and I hope this helps you understand the above conversation.

The different fields work as “stops” by selecting rays that comes from the object (the stop).  When we reduce the size of the stop, we get the performance as on-axis field at different “pupil” coordinate. For rays that forms the on-axis fields, they become the chief rays from the viewpoint of pupil imaging system. For systems whose aberrations are not well corrected, it’s not surprising to have large position differences between the center field and the maximum field if the stop size is large enough, which is the difference between the ENPP and the one traced by the new feature.

The method used to determine the pupil position is similar to the paraxial ray tracing method from my viewpoint. Using rays from different orientation instead of tangential and sagittal is similar to the method we use in spot diagram.

(Feel free to correct me if I am wrong.)