refactor: introduce parametrised nplike types

[agoose77]

The existing type hints for nplike were watered down to support a single ArrayLike protocol. This prevents us from imposing the constraint that a CuPy array can only be handled by a CuPy nplike.

This PR restores the original parametrised-nplike protocol, so that it's possible to declare an nplike as operating upon fixed array types. Unfortunately, mypy doesn't support higher-kinded-types, which are required to compose generic types. This means that we can't e.g. have

class NumpyArray(Generic[Backend]):
    @property
    def backend(self) -> Backend:
        ...

and so on. There are higher-kinded-type works related to mypy, e.g. https://sobolevn.me/2020/10/higher-kinded-types-in-python, but they don't look like a good fit for our needs; they have some drawbacks and impose constraints on the runtime logic.

So, this PR won't have huge sweeping impact, but it will allow us to explicitly type an nplike as a typetracer nplike, e.g.

def is_typetracer_nplike(nplike: NumpyLike) -> TypeGuard[TypeTracer]:
    return not nplike.known_data


if is_typetracer_nplike(nplike):
    print(nplike.zeros(4).report)

such that we can use fewer assertions.

[codecov]

Codecov Report

Merging #2795 (8049b9f) into main (c572295) will increase coverage by 0.01%.
The diff coverage is 87.21%.

Additional details and impacted files

Files	Coverage Δ
src/awkward/_backends/backend.py	88.09% <100.00%> (ø)
src/awkward/_backends/cupy.py	66.66% <100.00%> (ø)
src/awkward/_backends/dispatch.py	92.85% <100.00%> (ø)
src/awkward/_backends/jax.py	100.00% <100.00%> (ø)
src/awkward/_backends/numpy.py	100.00% <100.00%> (ø)
src/awkward/_backends/typetracer.py	78.12% <100.00%> (ø)
src/awkward/_broadcasting.py	95.43% <100.00%> (ø)
src/awkward/_categorical.py	58.62% <100.00%> (ø)
src/awkward/_connect/cling.py	25.71% <100.00%> (ø)
src/awkward/_connect/jax/reducers.py	86.20% <100.00%> (ø)
... and 143 more

[agoose77]

This is really just typing, so we should be safe to merge.