Do we want to additionally not include <any> in std.ixx when _HAS_STATIC_RTTI=0? It seems like we shouldn't be printing an error here.

@strega-nil-ms Agreed - the intention is for std.ixx to respect all of our various modes. I can validate and push changes.

Ok, I've pushed the following changes (FYI @CaseyCarter @strega-nil-ms):

• std.ixx: Guard <any> with _HAS_STATIC_RTTI.
  • <algorithm> is always included first, so <yvals_core.h> is available.
• Add /D_HAS_STATIC_RTTI=0 coverage to P2465R3_standard_library_modules.
  • Also reduce the /analyze coverage to one configuration. (It found no issues when originally added, so we don't need to spend an entire crosslist on it.) Combined, this reduces the number of configurations from 8 to 6.
• Guard __cpp_lib_any with _HAS_STATIC_RTTI, add test coverage.
  • We should have done this earlier (as we do for headers where /clr:pure is blocked, etc.) but now is also a good time.
• VSO_0000000_has_static_rtti: Comment the apparently unused include.

Thanks @frederick-vs-ja for working on this - if we're going to have this not-exactly-encouraged-but-still-supported mode, it should behave reasonably, and I think your change makes the story coherent. 😻

✅ Note to self: modules changes verified.

I'm mirroring this to the MSVC-internal repo - please notify me if any further changes are pushed.

Thanks for fixing the first bug after Standard Library Modules was merged! 🐞 ✅ 😻