Thanks for taking the time to capture and share the frames
I think the biggest issue there is that the images are very faint - without applying a stretch it is only really possible to see the one brightest star by eye (you can spot a few more if you zoom in and scroll around carefully). In fact, unless I use the option to apply the display stretch parameters during star detection, I struggle to reach the 18 detected stars that you see in the in focus frames. Ideally, somewhat longer exposures would bring more signal from the stars, a better signal to noise ratio and would make star detection work better.
Setting that aside, I do see the same pattern as you - that is that more stars are detected in the slightly out-of-focus frames than in the in-focus ones. I need to dig into that a bit more to see if I can work out why and if it is anything that causes concern in terms of the accuracy of focus.
cheers,
Robin