Pete,
That collimation error really needs to come down. Try magnifying the cross hairs in ccdins to get a better look at whats happening. Also, Steve67's (username) method,
from here, of collimation is really accurite i found aswell. I just got the shadow right in the middle of the defocused star, using the centre star on the chip, and collimation was sorted, even CCDINS agreed.
Yes, the aspect ratio measures the out of roundness of stars, The lower the % the better.
It may not be a tilt of collimation issue, but collimation or focuser/camera flexure could still change depending on the rigidity of yor settup and the location in the sky your pointing to. But if your getting significantly elongated stars, all over the chip its unlikely to be either of these.
HTH,
Josh