[FEA]: Expose get_cuda_native_handle to Python

[leofang]

Is this a duplicate?

• I confirmed there appear to be no duplicate issues for this request and that I agree to the Code of Conduct

Area

cuda.bindings

Is your feature request related to a problem? Please describe.

In #463 we added experimental C++ APIs get_cuda_native_handle for C/C++/Cython to access the C handles wrapped by cuda.bindings types/objects. This issue is to track the need of exposing such APIs to Python to replace the current usage of __int__(), something along this line:

from cuda.bindings.utils import get_cuda_native_handle
...

err, s = driver.cuStreamCreate()
s_ptr = get_cuda_native_handle(s)
assert isinstance(s_ptr, int)

We may also consider moving the C++ counterparts from cuda.core to cuda.bindings.

Describe the solution you'd like

n/a

Describe alternatives you've considered

n/a

Additional context

n/a