Is that the thing where after a number of reductions by a single process the BEAM will make sure another process gets its time in the sun? I believe so but I'm not familiar enough with the internals. They definitely spoke about reductions in the meeting I half-listened to.

Correct. It makes it so for(true) in other languages doesn't hijack the cpu and strive for the low latency goals of Erlang.

To be fair you can go about it by just filling up a process mail box with tons of messages.

Yes, Lumen has a scheduler and does preemption the same way the BEAM does.

Yes - preemptive multitasking is core to the BEAM process model.