There are some outliers, but as I mentioned before, I didn't yet do any quality control on the data. There were some (earlier) ones with very few stars and rather dodgy looking curves. And some later ones sometimes start with the focuser at a position quite far from focus due to temperature changes from the last run, resulting in a poor curve. I need to accumulate more data and apply some basic QC methods.
Thank you for following up on this. This looks very promising. I do recall a discussion on CloudyNights (CN) where they considered using the DEC guiding value as an indication of 'seeing.' Their reasoning was that if you had a decent polar alignment then there should be no DEC corrections other than whatever was due to star motion from seeing. So if you subtracted any drift in DEC then the resultant DEC rms motion should be also correlated with seeing. I have not more information than this, but I will try to fine the reference from CN.