Hi Roger,
What you are doing is the "Gemini Way" by using the drift method. The Gemini Polar Align Assist is simply an automated drift method process. Once you get close, you can optionally use the Polar Axis Correction (PAC) feature to get you a little closer. Don't expect to be able to use PAC straight away when you've just set up.
Predominately you should focus on the A and E figures. The other measurements can be used to optimise your set up further. However considering you are setting up and tearing down every night, there are a lot of changing variables. It's different for a permanent set up. I've spent some time manipulating these figures like "CF" by moving the counterweights further up the shaft while still maintaining good balance. Moving the worm block for FR and FD. There are many resources available on the net to optimise losmandy mounts.
I assume that every time you run PAC and make an elevation or azimuth adjustment you know that you must redo the pointing model. Straight after you made the changes, just do a cold start and start a new pointing model. Once you have a new pointing model, you can run PAC again to get closer if desired. I've ran PAC three times to average out seeing, rebuilding an accurate pointing model each time. PAC is only as accurate as your pointing model, which brings me to an early post you made. You mentioned that when you slew and center a start for alignment, the mount appears to over shoot the mark when pushing the controller button. It kind of drifts and looks like there is a lot of play in the system. I experienced this too, but only when in visual mode. I would highly recommend you change to photo mode when building the point model. This should resolve what you are experiencing. Alternatively use "all speeds mode". Though I've only used this mode in v3.x, not 4.x.
I had a look at the photo you posted. Is this raw, i.e. no post processing other the dark frame substracted? The stars look reasonably round to me, even when magnified.
I think you're on the right track.
