I think I can add clarity now to this discussion.
Andy Galasso (of PHD2) has been kindly discussing all of this with me for which I'm really grateful. As far as scope separation or parallax error he said it isn't a factor and presented this argument:
"we can forget about the separation since arcsin(15"/10ly)=0
"
However, I know from following Patrick Wallace's discussions of T-Point over at the Software Bisque forum that all polar alignment is a compromise and that field rotation can only be minimized. One only need look at the T-Point PA recommendations in the advanced settings to see how to compensate with ME or MA adjustments to get a PA that suits a particular need. So, I'm left with the conclusion that all fields rotate and it's just a question of how much one can tolerate.
Andy confirmed the following.
1. When a guide star is selected the guide star becomes the center point for the rotating field.
2. Multistar guiding forces a compromised center of rotation where the center is actually the mean position of all the guide stars.
a. So my question of looking at two different guide stars through a single scope (one guiding, one not guiding) will give the same result (except for DF!) as using two telescopes. The camera not guiding will detect the field rotation exactly as I did in my first post of this thread.
3. My own observation from #1 + #2 is that people with wide field guiders such as an ONAG + SX Ultrastar or ATIK 440EX need to be careful to select the same guide star from night to night! If this isn't done there will be multiple sets of images that have different centers of rotation. Registration might be compromised! Notice in my first experiment that the field rotated up to 1 - 1.5 pix in 5 minutes. Longer exposures would be even more problematic.
4. I was able to experiment further last night by guiding through one scope and just watching the same star through scope #2. When I first did this experiment I only had a sample of 1 successful image (of the guiding graph in PHD2) that I posted earlier. I speculated incorrectly that I couldn't believe that everything I was observing was DF. After repeating this in different areas I think I was quite wrong as I see totally different results for different areas along with greater deviations exactly where one might expect them....such as when the mount would be tracking west at around 45 degrees altitude. In this position gravity would be acting more and more on the focuser/camera combination. The attached photo demonstrates this quite well. The SX AO was able to successfully eliminate the DF in a second run at the same position.
If there was a mystery in my mind I think it is all nicely resolved!
Peter