[clang] [CIR][AMDGPU] Lower Language specific address spaces and implement AMDGPU target (PR #179084)

[David Rivera via cfe-commits]

RiverDave wrote:

> This PR is pre-requisite for amdgpu builtins codegen upstreaming. Thanks for working on this @RiverDave

Yes, I'll merge once ci's pass given I finally got the approval I wanted :D 

https://github.com/llvm/llvm-project/pull/179084