Doesn't the "spooky action at a distance" become moot if we set aside the notion of a single forward direction of time for information? That is, when you observe one half of the entangled pair there is not a super-luminal speed transfer of information because the information was already available to the other half of the pair at the time that they were entangled; it will become the necessary state upon observation of its pair because that is what is necessary in the future.
It's more complicated than that (in particular, you need to deal with conservation of information, which is difficult even in a classical particle physics model, and I don't know of any mathematical formalisms that make the quantum case any better), but very approximately yes (in roughly the same sense that, if you cut up a piece of cheese into smaller and smaller pieces, you eventually end up with a single cheese atom).
For example, if a photon decays (spontaneous pair production) into a electron and a positron, those two particles will have some of their information content 'entangled', and that is roughly the same information that would be 'preserved' accross a electron (say, in a atom) emitting a photon to drop into a lower energy level. Because those are the same particle interaction, rotated into different reference frames. (In practice, I think you mostly end up with entanglements like Bit_γ xor Bit_e xor Bit_e̅ = 0, which is why a electron and positron (or matter and antimatter in general) can always annihilate to form a photon (respectively photons) in yet another rotation of said particle interaction.)
Aren't you still discussing it in the form of a linear arrow of time? That initial decay is also the annihilation that forms a photon, when "reversed".
Setting aside the single direction of time seems even spookier than action at a distance. Is it really possible that when two atoms interact with each other, the result is partially determined by my decision two minutes from now about what way I’m going to measure the result of that interaction?
Maybe? The concept is called "retrocausality", though I find the name itself to be poorly chosen, since without a single direction of time there isn't really a "before" to be retrocausal towards.
Aside from whether it’s possible, it seems inevitable. Light does not experience time. There is no time between the emission and absorption of a photon.
