Yeah it ain't the easiest of decisions. @niko, when I think about the staged approach, it feels like it is maybe more disruptive than doing it all at once.
stage 1: force everyone to install k8s and celery libs -- that's one disruption / risk stage 2: remove pre-install -- another disruption it seems maybe better to do it all at once (if it's the ultimate goal). i suppose you could try to introduce a special deprecation warning after stage 1 if user is using celery without having installed the package explicitly. we might have to do some shenanigans to accomplish that but it might be possible. but even then, it just seems like more complexity and noise, prolonging the inevitable pain to me.
