[llvm] [CodeGen][AMDGPU] Remove premature empty subrange elimination (PR #201263)

[Matt Arsenault via llvm-commits]

================
@@ -0,0 +1,1524 @@
+; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 6
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a -stop-after=greedy,2 < %s | FileCheck %s
----------------
arsenm wrote:

Probably should run to the end. A mir input would be good but this is probably hard to reproduce 

https://github.com/llvm/llvm-project/pull/201263