Why doesn't gravity mess up the double slit experiment?

Yes, everything is a detector, but you need to quantify which things your system interacts with (and how strongly). Gravity is in some sense a poor example, because the quantum details of gravity are still an unsettled question (and gravity is a weak force regardless), so let's bypass that red-herring by replacing gravity with the electromagnetic field:

As your charged electron accelerates one way or another in a Stern-Gerlack apparatus or double-slit experiment, in theory it should radiate electromagnetic waves. Moreover, you would expect to be able to determine its position, by measuring differences of how particles in the environment are affected by the electron's EM field, right?

Basically, the reason you still observe interference fringes is because the coupling with the environment is weak. (Whereas if you gradually adjust the experimental parameters to increase the strength of coupling to the environment, then the fringes gradually fade.) Weak means that if you do the math, it isn't possible even in principle to infer sufficient information from the environment.

You might enjoy some of Zeilinger's journal-papers such as experimental demonstration of slit interference fringes with buckyballs (which are over a million times more massive than electrons), including demonstration of gradual decoherence (controlling the strength of interaction with the environment). You could also look at QM papers on weak measurement, or decoherence theory.