In classical terms, the difference between "redundancy" and "collapse", as you put it, would be, I suppose, the same as the difference between probabilities and statistics, with the latter computed from accumulated datasets. It seems to me that the "accumulated datasets", whether from experiments we consider to be "quantum" or not, are very close to objective, indeed for most physicists such datasets would be taken to be objective, insofar as everyone would see the same numbers on the screen as they scrolled through a dataset. Indeed, a classical collection of datasets is only distinguishable from a quantum collection of datasets by paying close attention (to what joint probability distributions can be constructed, as a particular example).
As you put it, only with provocations, putting the macroscopic numbers of degrees of freedom of "detectors" in the way of a prepared quantum state, does redundancy become collapse (although I think philosophers might phrase the same idea as potential becoming actual).
