I still want to pick up an a7sII. We mounted that to a telescope and cranked up the ISO. We could identify DSO as the telescope slewed to the next subject. It made framing super simple. Once framed, we'd drop the ISO back to realistic numbers for the actual image to be taken. With my DSLR, I'd have to take a 15 sec exposure just to see what I was looking at, and rinse-repeat for each adjustment.
The fact that the sensor is on for about 95% of the time through that process means that the heat isn't any different than a mirrorless setup.
Dark frame subtraction corrects reasonably well for this anyway. Cooled camera would perform better but that's a lot more trouble than it's worth for me as a hobbyist. I wouldn't recommend cooling with ice packs though. You want your sensor at a constant temperature through most of the data collection so that your dark frames actually work. If the temperature of the sensor fluctuates, your dark frames become not representative of the actual dark currents most of the time.
That is pretty much the only thing that I have found about ML that I don't like. Everything else about it makes my camera a much better system. ETTR is my fave!
It is admittedly hard to get the intervalometer to switch off though, often it keeps restarting itself after several restarts of the camera but eventually gives up.
Not really. I do a lot of astrophotography with a DSLR, but use the DSLR effectively as a mirrorless camera. I also regularly wedge in filters that prevent use of the SLR mechanism anyway. Considering my total exposure times are regularly in excess of 2 hours, heating isn't really any different from mirrorless, and mirrorless would be a blessing for its versatility and simplicity.
The only real way to combat heat issues for exposures that long is a cooled sensor.