Review reference docs for proper use of "reject"

In #70 (closed) @nickm proposes a technical use of the word reject. Specifically, we'll use "reject" when we mean that something not compiling is a thing the programmer (often, the template author) can rely on. When we just mean "this doesn't mean anything now but we make it work in the future" we'll say "isn't suported" or something.

We should review the docs and make sure the usage is consistent with this interpretation.

I'm going to file two tickets for this: this one is about the reference docs (including the rustdocs for the macros).