Everything is observer-dependent, no?
No, actually very little is. In particular, the dynamics of QM are essentially deterministic, both in the Schrödinger picture where one evolves the wavefunction forward in time and in the Heisenberg picture where one absorbs time as a parameter into the operators themselves (using, at least principally, unitarity to evolve what is now the operators acting on the state forward in time). The governing laws (symmetries, conservations, dynamics, etc.) are specific to the quantization scheme used, but the preference for one is a matter of convenience. And the are all dependent upon space and time even in terms of the formalism (perhaps if one uses the wavefunction as the dynamical object), so the wavefunction cannot be more fundamental then space and time which are required to define it even abstractly.
In fact, I suspect that it is the fact that the wavefunction is almost completely a classical system governed by classical, deterministic laws that makes it take center stage in so many explications, discussions, and so forth both for the layperson and among physicists working in e.g., quantum foundations (or even just writing textbooks or lecture notes). In the Schrödinger picture, only when one wishes to extract contextual information from the wavefunction formalism (along with the operators appropriate to the experimental situation at hand) does one need apply a different form of dynamical evolution that generally involves what is variously called a "collapse", "jump", "projection", etc. And in those interpretations of QM in which the wavefunction is most central, such as the many-worlds interpretation, there is no observer dependence because the wavefunction never collapses. In the de Broglie–Bohm approach, again there is no collapse but now two elements of the theory take on ontological meaning. In this approach, quantum mechanics (loosely speaking) concerns particle dynamics but the wavefunction exists independently, guiding the "particle" and remaining even after the "particle" is absorbed or otherwise registered via some apparatus (or naturally, all the time, during all physical interactions). Again, there is no observer dependence.
In the orthodox interpretation(s), and newer forms of it such as QBism, subjectivity plays a major role it is true. But in this case the wavefunction is not fundamental nor need it describe anything in reality at all. Generally speaking it encodes statistical information about the likelihood of experimental outcomes. That is, given a particular experimental design, systems prepared in a specified manner are then measured in a particular manner and the wavefunction encodes statistical information about the manner of specification (in particular, the degrees of freedom of the system).
But quantum mechanics cannot be fundamental as it is non-relativistic. Nor can it be made relativistic simply by attempting to "upgrade" the background arena to Minkowski space and demanding the system obey the requisite symmetries. The combination of non-commutativity of operators from quantum mechanics along with the energy-mass equivalence from relativity together (along with the demand that time be treated on similar footing as space) mean that any attempt at relativistic quantum
mechanics will fail. This is where the Bohmian approach, for example, runs into extreme difficulty and possibly other interpretations as well. What is required is QFT, which allows for the creation and annihilation operators (or raising and lowering) as well as a way in which to demote the spatial operators to be on the same footing as time, promoting the operators from QM to the status of systems state which ultimately act upon the vacuum.
So, in general, the Hamiltonian formulation in which QM is rooted is replaced by the simpler, scalar-valued Lagrangian familiar from classical field theory. As in classical field theory, the use of fields in quantum theory ensures causal constraints and locality while also giving us CPT and spin statistics along with a way in which to allow for relativistic interactions (e.g., the exchange of photons by charged particles in QED satisfying the U(1) symmetries, where the covariant derivative is introduced into the interaction to satisfy invariance without the need for symmetry breaking arguments as in the electroweak sector because like gluons photons are massive and unlike the QCD lagrangian, we are dealing with an abelian group).
But regardless, the contextuality of QM is a matter of considerable debate. It is true that in general there is no easy way to understand the dynamics of the theory in full without introducing something additional that isn't found in classical physics, whether it be the branching universes of MWI, abandoning locality, or abandoning the idea that the physical systems described in QM have the same kind of ontological properties as in classical physics.
In none of these approaches, interpretations, or extensions do we find somehow that entanglement is supernatural or that large systems can be causally described using terms borrowed from quantum theory but devoid of any actual meaning as you use them in your OP and elsewhere.
Learn some math and then some actual QM.
^^ This.