I still say that what is "weird" about quantum mechanics is that it is the place where we encounter the issues that Bohr was always talking about-- we can no longer pretend the physicist is a "fly on the wall." And it's not just that the measurement affects the system, we can treat that as little random perturbations that create normal measurement uncertainty. Instead, it is that the very process of
creating language about what is happening
requires the way measurement affects the system, that's what is new. We cannot simply imagine more and more precise measurements that cause smaller and smaller effects-- the effects are
fundamental, not to nature herself, but to
physics. We
need the effects, we need collapse, because collapse is just how we do physics-- we create the collapse on
purpose, it is not some kind of accidental or unfortunate side effect of a measurement. It is the effects that allow us to talk about what is happening, so we can never talk about what is
actually happening as if it was absent of those effects-- as if it would have happened even if we hadn't measured it. The interaction is what allows us to say anything about physical reality, so is
part of quantum mechanics. Bohr said as much in many ways.
As to whether or not a full system, including the physicist, remains in a pure state, that is not at all known (but is known to not be what the physicist perceives). It is a matter of interpretation. It's true that one can hold that all time evolution is unitary, so you only get a mixed state when you project out the things you don't care about, but this view has never been established as true (that's essentially the many worlds interpretation). The Copenhagen interpretation says that the rationalistic logic is backward-- we don't infer that all evolution is unitary because it makes sense to say that quantum systems evolve that way and everything else comprises of quantum systems, instead what we call "a quantum system" is
already a construct of our interaction with it, so collapse is there even
before we have a language about unitarity. So we can't actually say what is "the state" of the full system, be it mixed or pure, because we can't test it-- it ends up being whichever way we think physics works (stemming from the observer and the collapse, toward the quantum system, which is an empiricist approach, or stemming from the quantum system and culminating in the observation, which is a rationalist approach).
Fortunately, the predictions work the same either way, so we needn't declare our metaphysical bent before we start a calculation (and "shut up and calculate" also works). But we do have to make that declaration before we can talk about fundamentally ontological entities, like the state of the "whole system", that we do not as yet have any empirical evidence about. And to those who naively claim that decoherence resolves the issue, because the "whole system" can be shown to be uncollapsed, the Copenhagenist simply responds that you still don't know anything about the whole system until you observe it, which either makes it part of an even larger system, or involves the perception of a mind, whose functioning is quite unknown.
Finally, I don't agree that what is weird is what is certain or already determined. The EPR paradox is no issue if the state of both particles is determined, that's like the left and right socks in a pair, there's no problem with nonlocality unless the states are inherently indeterminate-- but indeterminate in a way that shows correlations that are impossible with local realism. Hence the weirdness of QM stems from the role of fundamental indeterminacy-- the lesson seems to be that if you structure physics to be about what a physicist can say about reality by interacting with it, then you discover you are forced to either conclude that reality is fundamentally indeterminate about certain questions in the absence of those interactions, or invoke additional unobservable elements (like pilot waves) that can seem like a magic invented for no other purpose than to relieve the mental burden of imagining inherent indeterminacy. What we must recognize is that none of the
interpretations of quantum mechanics, not Bohr's, not Everett's, not Bohm's, can
both give a coherent account of what happens in a measurement,
and explain why the physicist perceives only one outcome,
without invoking essentially magical effects that are inherently unobservable.
That's what is weird.