OK, last night it finally cleared up and I had the opportunity to test Eric's new scheduler (I presume that is included in the latest nightly build).
I only imaged two targets last night, the Rosette Nebula and M81/82. The scheduler performed flawlessly on these two sequential targets. My only suggestion for improvement would be to include a line in the code to the effect that
IF [next target >1hr into the future] THEN [park mount] ELSE [do nothing]
Currently, the mount continued tracking the target, although the next execution date was 16 hours in the future. Not a big deal for a mount in the backyard, but for a remote observatory it may lead to the telescope crashing into the tripod or pier.
One other thing one has to keep in mind is that the scheduler will NOT start executing the sequence, even if set to ASAP, IF a lead time > 0 is set AND the altitude limit set for the first target has already been reached. Lead time has to be set to 0 in that case.
The only other quirk I found relates to the exposure settings in the guide module. There, something strange happened. As I was testing the guide module before starting the scheduler on another target, I noticed that the image was blown out despite using a 0 gain setting and the same settings I had successfully used for months (3 s integration time, 1x1 binning). Strangely, SHORTENING the integration time did not yield a picture that was not white, i.e. overexposed, but EXTENDING the exposure time and INCREASING the binning to 2x2 did!?!? However, when I started the scheduler, with the same settings I had previously established optimal exposure for my guide cam image, now the image was blown out and I had to DECREASE the integration time again (to 3s) and the binning to 1x1. ????
After that, the system performed without a hitch for the rest of the night on a Pi3. Even solving was reasonably fast at ~20s.
Here the Dropbox link to last nights log: www.dropbox.com/sh/pints7iwug9r8b8/AAAXH...UOHb2wp33rNBpFa?dl=0
Just in case you need it. The log file was too large to attach.
All in all, the system performed very well.
Thanks to everyone who is putting countless hours into perfecting this!