[llvm] [CodeGen] computeRegisterProperties - fix `bf16` to `i16` bitcast quieting NaNs (PR #224867)

[Simon Pilgrim via llvm-commits]

================
@@ -571,6 +571,12 @@ class LLVM_ABI TargetLoweringBase {
   // values are always passed and returned as f32.
   virtual bool useFPRegsForHalfType() const { return false; }
----------------
RKSimon wrote:

Can we consistently use CConv for bf16 instead of adding useFPRegsForBFloat16Type? Even if ARMTargetLowering::getRegisterTypeForCallingConv isn't fully operational

https://github.com/llvm/llvm-project/pull/224867