I agree with most of the suggestions, but the "parameter-type-list" actually better captures the processing that occurs between the parameter-declaration-clause and the resulting types. (Thus, [dcl.fct] should be fixed.)

Oh, and "cv" is an established term for a set of cv-qualifiers, so I think we're good in that regard.

Is it actually necessary to repeat the full function type spelling, instead of just saying

if E1.E2 refers to a non-static member function and the type of E2 is FT…

I'm actually wondering about something else: Do we have to include "cv" and "ref-qualifier" in the resulting type for E1.E2? After all, those things only matter when checking the relationship of E1 and E2, but once that has been cleared, they're irrelevant for further processing. The relevant overload resolution section talks about "the non-static member function", so don't actually refer to "E1.E2" which is a funny prvalue invented for recursive parsing mechanics but not much else.