[llvm] [llubi] Upstream existing floating-point intrinsics (PR #196034)

[Matt Arsenault via llvm-commits]

================
@@ -84,6 +84,11 @@ static cl::opt<bool>
                   cl::desc("Disable interpreter-introduced non-determinism."),
                   cl::init(false), cl::cat(InterpreterCategory));
 
+static cl::opt<bool>
+    FuseMultiplyAdd("fuse-multiply-add",
----------------
arsenm wrote:

Name and description should elaborate this is just about fmuladd. The default should also be true 

https://github.com/llvm/llvm-project/pull/196034