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

[Matt Arsenault via llvm-commits]

================
@@ -847,13 +849,15 @@ class RcpPat<Instruction RcpInst, ValueType vt> : AMDGPUPat <
 // Instructions which select to the same v_min_f*
 def fminnum_like : PatFrags<(ops node:$src0, node:$src1),
   [(fminnum_ieee node:$src0, node:$src1),
-   (fminnum node:$src0, node:$src1)]
+   (fminnum node:$src0, node:$src1),
+   (fminimumnum node:$src0, node:$src1)]
----------------
arsenm wrote:

It would help do this in baby steps, by first introducing fminimumnum/fmaximumnum as legal without all this other stuff 

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