I am reporting this using v2.5.2.2
This new behaviour does not seem to function as you expected. There is a good chance that this applies to others as well, but unless you are looking for the focus position in the FITS header you will not notice if your flat was taken at the wrong focus as it is almost impossible to notice if a flat is out of focus unless you are having flats issues in image calibration.
What is happening is...
Focus with the current filter, e.g red in this case. Take the images, then take the flats. If the flat your taking is the same as the one currently imaged with and it's in focus, no problem the focus is already set and the flats will be taken ok. But if you then have another event to take flats with a different filter. SGP takes the 'absolute' focus position from the offsets in the filter list. So it actually just goes to what ever figure is listed there and does not alter the focus by the offset as it used to.
If you have further events with other flats, they will also be wrong and if you have more lights to take after those flat event, the start filter position will be wrong because it then performs the normal offset, but as it starts of at the wrong place it will end up in the wrong position. So if you have not got the control panel set to auto focus on a filter change. All subsequent images will be out of focus.
As a get around, (and this only works if you can manually edit each event in real time) is to just do one event line for a flats set for the filter currently in use. Then manually move to the next filter you want to take flats for, then start that flats event. That way the focus will be where you expect it to be.
To help I have a log for the start of a session. I aborted the flats event when I realised what was going on. Unfortunately SGP hung on the abort! so I had to close it.
I do hope I have explained this well enough, just ask me to clarify if not. This happened on my last session and I had to re-take the flats. But then I was not sure if I had done something wrong. This time however I realised in time.
Log file:
http://www.mikeoates.org/sgp/sg_logfile_20161005174350.txt
Mike