[clang] [clang] Add support for [[clang::diagnose_if(...)]] and [[clang::enable_if(...)]] syntax (PR #225900)

[via cfe-commits]

higher-performance wrote:

FWIW I'm not sure I see a problem with just supporting it after the parameter list, either. I've been taking it as a given that it is a problem, but I'm not sure I see it. Does C++ standard require the function type to become different by mere virtue of an attribute being placed there? Could it possibly even do that in theory, given that it cannot possibly dictate the meanings of the attributes or dictate when two types would in fact end up identical? Semantically it's clear what the attribute in that position should do, so I'm not sure where exactly the problem lies.

https://github.com/llvm/llvm-project/pull/225900