Add support for std::this_thread::sleep_for

There seems to be an issue with `chrono::duration` being host only.

` std::this_thread::sleep_for(std::chrono::microseconds(1));`

generates the following compilation error:

`error: calling a constexpr __host__ function("duration") from a __device__ function("GetEventIndex") is not allowed. The experimental flag '--expt-relaxed-constexpr' can be used to allow this.`

Trying to replace `std::chrono` by `cuda::std::chrono` causes a different compilation error

`error: no instance of function template "std::this_thread::sleep_for" matches the argument list
argument types are: (cuda::std::__3::chrono::microseconds)`