We do know that they can't disagree as collapsed particles behave differently than uncollapsed particles.
> It's just that one observer is unsure of what result they will get for a subsequent experiment, while the other observer can know it exactly.
Quantum systems aren't about knowledge or statistics, the wave function is an actual physical thing that changes how the particle behaves. If the wave function has collapsed it no longer behaves the same as before, so what you said here is wrong.
You seem to just want to argue for the sake of arguing.
I imagine you're thinking of the version of the double slit experiment with a detector installed next to one of the slits, which prevents the interference pattern from forming. However, this only applies for a "measurement device" installed at the slits. That is, a classical system interacting with the photon/electron/etc before it hits the classical screen. But this is not mysterious at all: the classical device causes decoherence of the wave function and thus prevents self-interference of the beams traveling through the two slots. That is true even in interpretations that consider that wave function collapse is not a physical phenomenon, such as Many Worlds.
But this doesn't prove much. We already know that a photon hitting an atom doesn't cause wave function collapse, while a photon hitting a detector does. We can say that an atom is not a measurement device and an observer is, but this immediately raises the question of why. Or, we can choose the MWI route and say that the wave function never collapses, and the only difference between an atom and a detector is that we are not entangled with that atom, but we are entangled with the detector. And if so, it immediately follows that if we could get a detector that we are not entangled with, we would not see the same results as we do in the traditional double slit + detector on a slit experiment.
For example, when performing measurements on two entangled particles at very far away places, the wave function collapses instantly, across any distance, even light years away. After that experiment, the result on the other side is 100% determined (if performed in the same basis). But there is no experiment whatsoever that could be done to tell if the collapse has happened or not from the other side.
This is the resolution of the famous EPR paradox: wave function collapse is non-local (instantaneous), but it is probably impossible to send classical information with it faster than the speed of light limit.
Do note that the result of the experiments is not going to differ, it is guaranteed to be consistent by the fact that both experiments are measuring the same quantum system. It's just that one observer is unsure of what result they will get for a subsequent experiment, while the other observer can know it exactly.
And in the many worlds interpretation, it is indeed believed that apparent wave function collapse is a relative phenomenon, caused by entanglement with the classical environment, which spreads out at the speed of light.