EPIC: Typing enhancement

[leofang]

cuda.core can benefit from clean and detailed type hints. The typing annotation can guide users to understand the API design and expected workflow.

[carterbox]

Something that I noticed when using migrating nvmath-python to cuda.core is that not all types are importable from the same level. For example

from cuda.core.experimental import Device, Stream
from cuda.core.experimental._event import Event # shouldn't need to import _event to get Event

def some_function(device: Device, stream: Stream, event: Event) -> None:
 pass

If you're a downstream project and you also want to use type hints, the Event type is not part of the public API, but it is returned from parts of the public API, so it should be accessible without having to import a private module.

[mdboom]

I think #2061 largely closes this bug, with the exception of the subissue #454 (stop repeating types in docstrings).