[llvm] [AMDGPU] Use `v_cvt_pk_*` instructions for i16_f32 saturated conversions (PR #202680)

[Jay Foad via llvm-commits]

================
@@ -3949,6 +3949,9 @@ SDValue AMDGPUTargetLowering::LowerFP_TO_INT_SAT(const SDValue Op,
   if (DstVT == MVT::i16 && SatWidth == DstWidth && SrcVT == MVT::f16)
     return Op;
 
+  if (DstVT == MVT::i16 && SatWidth == 16 && SrcVT == MVT::f32)
+    return Op;
----------------
jayfoad wrote:

```suggestion
  if (DstVT == MVT::i16 && SatWidth == DstWidth && (SrcVT == MVT::f16 || SrcVT == MVT::f32))
    return Op;
```

https://github.com/llvm/llvm-project/pull/202680