I collect flats each morning for each filter used during the night. I have a sequence file for that with 3 jobs for B/G/R filters configured as wall flats. I then add that sequence at the end of my schedule with target name set to date ("231110" in this case), so that the files end up in distinct folders. The B filter works fine. The rest, G and R, lose the target prefix and end up in the same folder every day. At least the counter is persistent so I don't lose yesterday's flats, but they all are in the same folder. (But it can't track the progress and keeps spinning in this job capturing more and more G and R flats until I stop it manually.) This behavior did not change in the last version. It was broken at least since 3.6.6, maybe even earlier.
See the logs. Scroll to circa 10:22:50. All the B images before that point had "seq prefix = 231110_Flat_B" and captured correctly. The G probe captures also have good looking prefix but as soon as it figured the right exposure and started saving images, it reverted to "seq prefix = Flat_G" and never went back. Same happens later for R.
I caught an even better one last night! I only shot H subs, so only one simple flats job at the end. But instead of having the '231115' prefix from the schedule's target name it borrowed the 'Rosette_11' prefix from the last executed lights job. This simple scenario used to work fine before and I had no updates last days. The only new thing last night was that my schedule has two groups of jobs cycling through each for a few hours.
Konstantin,
it took a while until I got your setup working - and I found a bug that needed to be fixed first. Meanwhile I understand the problem, but I cannot reproduce it. There have been changes to the logic of loading esq files, therefore I guess these changes have already fixed the problem that you experienced.
I would therefore recommend using the latest build and updating the flats esq files. Especially the H.esq file, which is the oldest one.