[clang] [clang] Add test for CWG156 "Name lookup for conversion functions" (PR #121654)

[Vlad Serebrennikov via cfe-commits]

Endilll wrote:

> I wonder if we just mark that as superseded and/or N/A.

To do that, we need to downgrade CWG1111 from being available in Clang 3.2 to "partial", and merge in at least the test that doesn't pass. Maybe we should do that anyway.

> The original requirements that lookup must find "the same thing" everywhere is gone so in effect we will never support the behavior that this issue is describing

Note that it's resolved as NAD, so it's totally fine that the behavior described in the issue is not what we test.

https://github.com/llvm/llvm-project/pull/121654