I was thinking of SQM from a unihedron sky quality meter. If there is one is connected to INDI then collect data from it. I think its probably very difficult to estimate real SQM (in units of MPSAS) from an image.
So basically if there is an SQM device and it is connected with INDI, collect that data every x min or at every capture.
Yes, I have that after focus correction As a work around I include a 10second wait in all my Ekos sequences. That 10 seconds is usually enough for the guiding to get back on track. I was thinking to try and not suspend guiding during focus next time and see how it goes.