[CIR][AMDGPU] Add module flags for AMDGPU target

[skc7]

This PR adds support to emit AMDGPU-specific module flags amdhsa_code_object_version and amdgpu_printf_kind to match OGCG behavior.

In CIRGenModule, the flags are stored as CIR module attributes:

• cir.amdhsa_code_object_version (integer)
• cir.amdgpu_printf_kind (string: "hostcall" or "buffered")

During lowering to LLVM IR (in LowerToLLVMIR.cpp), these attributes are converted to LLVM module flags.

[skc7]

amdhsa_code_object_version and amdgpu_printf_kind are llvm ir level module flags. Not sure if at CIR level does targets need a specific attribute for them implemented. So, went ahead with integer and string attribute representation.

CC: @bcardosolopes, @xlauko

[bcardosolopes]

Overall makes sense, some nits

[bcardosolopes]

Create a emitAMDGPUMetadata (similar to emitOpenCLMetadata) and add this code over there.

[skc7]

Thanks for feedback. Have updated PR to use emitAMDGPUMetadata() instead.

[bcardosolopes]

The name of these attributes shouldn't be hardcoded, please use something akin to getOpenCLVersionAttrName here and in the one below.

[skc7]

getAMDGPUCodeObjectVersionAttrName()
getAMDGPUPrintfKindAttrName()

above methods have been added for querying the attr names.