Fix/suppress a type warning in PyTorch (#55142)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/55142
Declare some functions C10_HOST_DEVICE to fix the NVCC warning.
During pytorch compilation, NVCC compiler was emmiting several warnings like this one:
```
caffe2/c10/util/TypeCast.h(39): warning: calling a constexpr __host__ function from a __host__ __device__ function is not allowed. The experimental flag '--expt-relaxed-constexpr' can be used to allow this.
detected during:
instantiation of "dest_t c10::static_cast_with_inter_type<dest_t, src_t>::apply(src_t) [with dest_t=c10::complex<double>, src_t=__nv_bool]"
(158): here
instantiation of "To c10::convert<To,From>(From) [with To=c10::complex<double>, From=__nv_bool]"
(170): here
instantiation of "To c10::checked_convert<To,From>(From, const char *) [with To=c10::complex<double>, From=__nv_bool]"
caffe2/c10/core/Scalar.h(63): here
```
How to reproduce.
- Make sure you are on remote/master
- run:
`buck build mode/dev-nosan caffe2/torch/fb/sparsenn:sparsenn_operators_gpu`
Test Plan: - compilation completes without warnings.
Reviewed By: r-barnes
Differential Revision: D27469757
fbshipit-source-id: f8c4eedb637c6d487ac49bb310e48be11db204e2
