[clang] [Clang] Add `__builtin_[ordered_]reduce_addf` for ordered/unordered fp reductions (PR #176160)

[Andy Kaylor via cfe-commits]

================
@@ -1136,6 +1136,7 @@ RValue CIRGenFunction::emitBuiltinExpr(const GlobalDecl &gd, unsigned builtinID,
   case Builtin::BI__builtin_reduce_xor:
   case Builtin::BI__builtin_reduce_or:
   case Builtin::BI__builtin_reduce_and:
+  case Builtin::BI__builtin_reduce_addf:
----------------
andykaylor wrote:

No, I'm saying the `f` suffix specifically refers to a variant that handles 32-bit floating point. For example, `BI__builtin_fma` handles 64-bit floating-point values and `BI__builtin_fmaf` handles 32-bit floating-point values (consistent with the naming convention used for libm functions).

https://github.com/llvm/llvm-project/pull/176160