[clang] [clang][Sema] Split SYCLKernel back into its own attribute (PR #162868)

[Joseph Huber via cfe-commits]

================
@@ -1599,8 +1599,15 @@ def CUDAShared : InheritableAttr {
 }
 def : MutualExclusions<[CUDAConstant, CUDAShared, HIPManaged]>;
 
+def SYCLKernel : InheritableAttr {
+  let Spellings = [Clang<"sycl_kernel">];
+  let Subjects = SubjectList<[FunctionTmpl]>;
+  let LangOpts = [SYCLDevice];
----------------
jhuber6 wrote:

You should just require `SYCL` and probably ignore it on the CPU, I think that's how CUDA's `__global__` works.

https://github.com/llvm/llvm-project/pull/162868