Add Pydantic model for cellpose channel input?

[jluethi]

The review of input arguments & the discussions on pydantic models for napari workflows (e.g. #404) and channel (#386 (comment)) made me think:

We should also refactor the way we handle the channel inputs in Cellpose.

Currently, we do:

wavelength_id: Optional[str] = None,
channel_label: Optional[str] = None,
wavelength_id_c2: Optional[str] = None,
channel_label_c2: Optional[str] = None,

But actually, this is very similar to the input specs of napari workflows, just with slightly different limits: we can only use "image" as type and only have up to 2 images, where order matters

=> let's see whether we can use a variant of what we build for napari workflows input & output arguments here as well!

[tcompa]

Ref #404 (comment)