[clang] [CUDA] Enable variadic argument support in front-end (PR #161305)

[via cfe-commits]

LuoYuanke wrote:

> There are some flags like `-fcuda-enable-variadic` that we pass in places. These should likely be deprecated as the variadic support on GPUs has been well tested for over a year now. Just keep in mind the ABI is different between AMDGPU and NVPTX.

I can help to remove the code for  sema check on variadic declaration and deprecate `-fcuda-enable-variadic` option with another patch. I don't have commit access. Could you help to land this patch?

https://github.com/llvm/llvm-project/pull/161305