[clang] [clang][Sema][CUDA] Restrict immediate template resolution to host-device functions (PR #200662)

[Yaxun Liu via cfe-commits]

yxsamliu wrote:

Thanks for reducing the test. I still do not see a conversion function template case here. `operator=` covers the method-template path, but `AddTemplateConversionCandidate` is for conversion functions such as `template <class T> operator T()`. Could you add one small case for that path, or point me to the part of this test that is meant to cover it?

https://github.com/llvm/llvm-project/pull/200662