[llvm] AMDGPU: Validate processor and features in TargetID parsing (PR #209304)

[Matt Arsenault via llvm-commits]

arsenm wrote:

The splitTargetID case is ultimately for a diagnostic I think shouldn't exist (getConflictTargetIDCombination). It forbids using gfx90a and gfx90a:xnack+ gfx90a:xnack- or in the same compile, but I don't really see the point. Particularly in the clang-offload-bundler usage. 

https://github.com/llvm/llvm-project/pull/209304