There is no spacetime model of many worlds at all, as far as I know. Many worlds as an interpretation of QM is based on non-relativistic QM.

The "space" in which quantum states "live" is not spacetime but Hilbert space. "Many worlds" is really a misnomer in the sense that there is only one quantum state in Hilbert space. But quantum states in Hilbert space can have relationships to ordinary space that are not at all like those in classical physics: for example, a single quantum state can describe a system containing multiple entangled particles that are spatially separated; in such a state no particle by itself has any definite state at all (which is impossible in classical physics). That is true regardless of which QM interpretation you adopt.

> we’re just exploring one probabilistic trace through it

There is no such thing as a "probabilistic trace" through a deterministic 4-dimensional spacetime in which all events are fixed in advance.

Even the idea that anything is being calculated is a supposition. There appear to be relationships between entities and some of them can be modelled using human math on human-accessible time scales.

But although some of the models have potentially infinite precision the universe itself clearly doesn't on any scale.

Or if it does, it's not accessible to us. So whatever is happening may use some other paradigm to manage relationships, and the entities and relationships we see are emergent and not at all fundamental.

this is also why I think that artificial consciousness (big difference to intelligence) is a whole lot harder than many people think. a stack of papers where someone has manually written the neural net (or whatever technology) computation and results is unlikely to have a subjective experience. because there is an indirection between a simulation and the real world. the simulation is a computation. the real world happens.