Could you verify I'm understanding this correctly? If I increase

datastore-revision-quantization-interval

significantly, minimizeLatency would be stale for up to that period, but using zed tokens with atLeastAsFresh consistency I could reduce the staleness based on my context?