Thanks! Do consequentialist kind of port the first axiom (completeness) from the VN-M utility theorem, changing it from decision theory to meta-ethics?

Not explicitly (except in the case of some utilitarians), but I don't think many would deny it. The boundaries between meta-ethics and normative ethics are vaguer than you'd think, but consequentialism is already sort of metaethical. The VMN theorem isn't explicitly discussed that often (many ethicists won't have heard of it), but the axioms are fairly intuitive anyway. However, although I don't know enough about weird forms of consequentialism to know if anyone's made a point of denying completeness, I wouldn't be that surprised if that position exists.

To put that in one popular context: in the Trolley Switch and Fat Man problem, it seems like most people start with the assumption that there exists a right answer (or preferable, or best, whatever your terminology), and that it could never be the case that an agent will do the wrong/immoral/unethical thing no matter what he or she chooses. Am I right that this assumption exists?

Yes, I think it certainly exists. I'm not sure if it's universal or not, but I haven't read a great deal on the subject yet, you I'm not sure if I would know.