Refactor backend classes to use generic type parameters and improve type hints across various backends

[david-zwicker]

No description provided.

[Copilot]

Pull request overview

This PR refactors the backend layer to use a generic native-array type parameter, updating concrete backends (NumPy/JAX/Torch) accordingly, and applies small typing improvements across PDE implementations and utility helpers.

Changes:

• Introduce BackendBase[TNativeArray] and update backend subclasses to specify their native array type (e.g., jax.Array, torch.Tensor, NumericArray).
• Improve type hints in several PDEs by explicitly typing operator-construction kwargs dictionaries.
• Adjust a few numerical/util functions with updated annotations and additional type: ignore suppressions.

Reviewed changes

Copilot reviewed 12 out of 12 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
pde/backends/base.py	Make backend base class generic and update operator return types to use TNativeArray.
pde/backends/numpy/backend.py	Parameterize NumpyBackend with NumericArray and adjust imports/signatures for typing.
pde/backends/jax/backend.py	Parameterize JaxBackend with jax.Array and remove some signature-level ignores.
pde/backends/torch/backend.py	Parameterize TorchBackend with torch.Tensor and remove some signature-level ignores.
pde/backends/numba/backend.py	Remove a # type: ignore on make_pde_rhs signature (typing alignment cleanup).
pde/solvers/base.py	Add a suppression on Euler step to accommodate new generic backend typing.
pde/tools/spectral.py	Minor numeric annotation tweaks and add suppression on FFT-domain scaling.
pde/tools/output.py	Add suppression on tqdm.auto import alias.
pde/pdes/swift_hohenberg.py	Type the operator kwargs dict as dict[str, Any].
pde/pdes/kuramoto_sivashinsky.py	Type the operator kwargs dict as dict[str, Any].
pde/pdes/kpz_interface.py	Type the operator kwargs dict as dict[str, Any].
pde/pdes/cahn_hilliard.py	Type the operator kwargs dict as dict[str, Any].

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Avoid using a blanket # type: ignore on the tqdm.auto import. The repo’s mypy config already sets ignore_missing_imports = true for tqdm.* and also enables warn_unused_ignores, so this can become an “unused ignore” warning (and mask real issues). Prefer removing the ignore; if a specific type-checker error remains, suppress it narrowly (e.g., by adjusting typing under TYPE_CHECKING or using a local cast).

Suggested change

	from tqdm.auto import tqdm as progress_bar_class # type: ignore
	from tqdm.auto import tqdm as progress_bar_class

[Copilot]

Please avoid introducing a blanket # type: ignore for the in-place multiply. With warn_unused_ignores = true, this can easily turn into an “unused ignore” once types are adjusted, and it also hides the underlying typing mismatch. Prefer making the types explicit here (e.g., assign the FFT result to a separately-typed variable / use a cast for the complex FFT result) so the operation type-checks without suppression.