[llvm] [AMDGPU][GlobalISel] Enable vector reductions (PR #131413)

[Matt Arsenault via llvm-commits]

================
@@ -426,6 +426,7 @@ class LegalizerHelper {
   LegalizeResult lowerMinMax(MachineInstr &MI);
   LegalizeResult lowerFCopySign(MachineInstr &MI);
   LegalizeResult lowerFMinNumMaxNum(MachineInstr &MI);
+  LegalizeResult lowerFMinimum_FMaximum(MachineInstr &MI);
----------------
arsenm wrote:

I'd keep the base expansion separate from the reduction expansions 

https://github.com/llvm/llvm-project/pull/131413