[AMDGPU] Revert Old optimization to fold READFIRSTLANEs across BBs#446

ronlieb merged 2 commits intoamd-stagingROCm/llvm-project:amd-stagingfrom

amd/dev/akadutta/remove-redundant-rfl-opt-revertROCm/llvm-project:amd/dev/akadutta/remove-redundant-rfl-opt-revertCopy head branch name to clipboard