[clang] [llvm] [AArch64][SME2] Refine fcvtu/fcvts/scvtf/ucvtf (PR #77947)

[Sander de Smalen via cfe-commits]

================
@@ -10056,7 +10056,7 @@ CodeGenFunction::getSVEOverloadTypes(const SVETypeFlags &TypeFlags,
 
   llvm::Type *DefaultType = getSVEType(TypeFlags);
 
-  if (TypeFlags.isOverloadWhile())
+  if (TypeFlags.isOverloadWhile() || TypeFlags.isOverloadMultiVecCvt())
----------------
sdesmalen-arm wrote:

Given that they have the same semantics, can we rename the flag so that we avoid having to allocate another bit? 

https://github.com/llvm/llvm-project/pull/77947