[clang] [llvm] [DirectX] Implement UseNativeLowPrecision shader flag analysis (PR #134288)

[Finn Plummer via llvm-commits]

================
@@ -282,10 +282,14 @@ void CGHLSLRuntime::addHLSLBufferLayoutType(const RecordType *StructType,
 
 void CGHLSLRuntime::finishCodeGen() {
   auto &TargetOpts = CGM.getTarget().getTargetOpts();
+  auto &LangOpts = CGM.getLangOpts();
   llvm::Module &M = CGM.getModule();
   Triple T(M.getTargetTriple());
   if (T.getArch() == Triple::ArchType::dxil)
     addDxilValVersion(TargetOpts.DxilValidatorVersion, M);
+  if (LangOpts.NativeHalfType)
----------------
inbelic wrote:

Can we add a comment here that describes why we can just use `NativeHalfType` to set this flag and the relationship that it has iwth `enable-16bit-types` in DXC

https://github.com/llvm/llvm-project/pull/134288