[libc] [llvm] [libc][math][c23] Add fmaf16 C23 math function. (PR #130757)

[via llvm-commits]

================
@@ -90,14 +90,8 @@ class FmaTestTemplate : public LIBC_NAMESPACE::testing::FEnvSafeTest {
     // Test overflow.
     OutType z = out.max_normal;
     InType in_z = LIBC_NAMESPACE::fputil::cast<InType>(out.max_normal);
-#if defined(LIBC_TYPES_HAS_FLOAT16) && !defined(__LIBC_USE_FLOAT16_CONVERSION)
-    // Rounding modes other than the default might not be usable with float16.
-    if constexpr (LIBC_NAMESPACE::cpp::is_same_v<OutType, float16>)
-      EXPECT_FP_EQ(OutType(0.75) * z, func(InType(1.75), in_z, -in_z));
-    else
-#endif
----------------
overmighty wrote:

Deleting this code means some includes are no longer needed:

```diff
-#include "src/__support/CPP/type_traits.h"
 #include "src/__support/FPUtil/cast.h"
-#include "src/__support/macros/properties/types.h"
```

https://github.com/llvm/llvm-project/pull/130757