Nit: consider giving `ADTFunction` type arguments

[pieter-bos]

Currently ADTFunction "inherits" the type arguments of its surrounding ADT, but this is inconvenient to maintain. This is because ADTFunction invocations currently have to maintain a reference to the ADT in order to map/type check the type arguments.

The drawback of this is that the straightforward encoding of this is to just give each function its own separate domain, which is kind of ugly, and inheriting the type arguments signals the usual case where the function signature is only dependent on the type arguments of the ADT it is related to.