I actually think the way it currently works is more likely to leave globs at the start of a new layer, because of the large travel needed. I see now why things were changed for multithreading, but perhaps an alternative could be to process layers in pairs, so each thread gets two concurrent layers to process. This would allow it to start and finish each pair of layers in the same spot. Either way, an option to toggle this behaviour would be nice, because I think most of the time there aren't any parts small enough to suffer from not being able to cool in time (at least certainly in the objects that I print). Perhaps even someone could point me to the area of code that would need to be changed?
I'm currently running v3.5.1 and are wondering the same thing. I would much prefer if it printed multiple objects like this: LAYER 1: 1, 2, 3, 4 LAYER 2: 4, 3, 2, 1 LAYER 3: 1, 2, 3, 4 LAYER 4: 4, 3, 2, 1 etc. The way it currently works, when there are lots of little objects on the build plate, there is a massive travel at the end of each layer, which is not very efficient.