[llvm] AMDGPU: Fix incorrectly selecting fp8/bf8 conversion intrinsics (PR #107291)

[Stanislav Mekhanoshin via llvm-commits]

================
@@ -1444,7 +1451,7 @@ def FeatureISAVersion9_4_Common : FeatureSet<
    FeaturePackedFP32Ops,
    FeatureMAIInsts,
    FeatureFP8Insts,
-   FeatureFP8ConversionInsts,
+   FeatureFP8ConversionInsts, FeatureCvtFP8VOP1Bug,
----------------
rampitec wrote:

Why not on a new line?

https://github.com/llvm/llvm-project/pull/107291