Technically it’s not really a horizon if it “opens up” allowing you to observe events from the inside afterwards. But of course in any realistic setting (including that experiment) it will open up eventually, so no horizon. But nature doesn’t know that it will open up, so maybe it should behave like a horizon until nature knows, resulting in a criteria like you said. I think the criteria is loosely equivalent to saying “the acceleration must change the speed by almost c”, so your centrifuge probably wouldn’t lead to radiation.

But I really am not sure about any of this. The right way to do this is to actually calculate the mode function. One day when I’m better with QFT and all these stuff I’ll try to do it.