Well I tried this evening and couldn't replicate the issue using some test scripts (on KStars version 3.5.4 Stable (Build 2021-07-15T21:15:09Z)).

I tried the following: (1) one target with 1000 images, and (2) 25 targets with 50 images each (for ease they were actually all the same target to be quick).

Thinking out loud the obvious variables which are different from last time (software versions put aside of course!) are (1) I did not load a .esl sequence file, I manually created it then ran it, (2) only one target used, with no focus etc and (3) a single .esq file used for each target.

I will have to wait for some better weather and test in anger as they say.