[llvm] [AMDGPU] Adjust amdgpu fmax/fmin legalization (PR #202917)

[Matt Arsenault via llvm-commits]

================
@@ -177,12 +177,13 @@ defm FmaMixBF16Insts : AMDGPUSubtargetFeature<"fma-mix-bf16-insts",
 >;
 
 defm IEEEMinimumMaximumInsts : AMDGPUSubtargetFeature<"ieee-minimum-maximum-insts",
-  "Has v_minimum/maximum_f16/f32/f64, v_minimummaximum/maximumminimum_f16/f32 and"
-  "v_pk_minimum/maximum_f16 instructions"
+  "Has v_minimum/maximum_f16/f32/f64, v_minimummaximum/maximumminimum_f16/f32,"
+  "v_pk_minmax/maxmin_num_f16, v_min/max_num_f16/f32/f64,"
+  "and v_pk_min/max_num_f16 instructions"
----------------
arsenm wrote:

Don't know why this is touching this but introducing typos 

https://github.com/llvm/llvm-project/pull/202917