That's not fair - plenty of people have defined "observer" in plenty of different ways (including defining it away, like decoherence supporters would prefer), it's just that there's no consensus because it's mostly a matter of philosophy without any experimental consequences.
As for nonlocality, entanglement is a fundamentally nonlocal feature of quantum mechanics. There's no way to excise that from the theory, even if you can't exploit it to transmit information faster than light. I'd say calling it inherently nonlocal is entirely fair, that's a very different claim than that we can transmit information faster than light.
...which, as an aside, would not even in and of itself necessarily be problematic, closed timelike loops (for instance) are perfectly fine and good as long as you account for the fact that quantum interference will have very strong effects even at a macro scale in such geometries (the grandfather paradox gets interfered away, more or less, even if macroscopically unlikely things need to happen to end up with a reinforced consistent timeline).
As for nonlocality, entanglement is a fundamentally nonlocal feature of quantum mechanics. There's no way to excise that from the theory, even if you can't exploit it to transmit information faster than light. I'd say calling it inherently nonlocal is entirely fair, that's a very different claim than that we can transmit information faster than light.
...which, as an aside, would not even in and of itself necessarily be problematic, closed timelike loops (for instance) are perfectly fine and good as long as you account for the fact that quantum interference will have very strong effects even at a macro scale in such geometries (the grandfather paradox gets interfered away, more or less, even if macroscopically unlikely things need to happen to end up with a reinforced consistent timeline).