Add static type annotations about dask HOT 4 OPEN

Hi @benrutter, I agree that we don't need to take on the whole repo and there is plenty of low hanging fruit. The errors I am running into are:

• da.from_array - chunks is inferred as a str
• array.rechunk - chunks is inferred as a str
• dd.from_delayed returns DataFrame | Series and will always require a cast. Not sure how to fix that other than adding helpers.
• map_blocks seems to not behave properly as a method in the eyes of pyright:

da.from_zarr(core.core_img).map_blocks(
    segment_tissue,
    dtype=np.uint8,
    drop_axis=0,
    downsample_factor=tissue_masks_params.downsample_factor,
)

Error:

Argument missing for parameter "func"

If you add explicit func=segment_tissue, then get an error about self:

Argument missing for parameter "self"

Perhaps something wacky with @wraps:

    @wraps(map_blocks)
    def map_blocks(self, func, *args, **kwargs):
        return map_blocks(func, self, *args, **kwargs)

map_blocks is of the following type. Looks like it is no longer recognized as a method or something.

map_blocks: _Wrapped[(func: Unknown, *args: Unknown, name: Unknown | None = None, token: Unknown | None = None, dtype: Unknown | None = None, chunks: Unknown | None = None, drop_axis: Unknown | None = None, new_axis: Unknown | None = None, enforce_ndim: bool = False, meta: Unknown | None = None, **kwargs: Unknown), Unknown, (self: Array, func: Unknown, *args: Unknown, **kwargs: Unknown), Unknown] | Any

from dask.

We are aware of the issue but there is currently no plan on addressing this from within the currently active contributors. We currently have other (higher) priorities to focus our effort on. Contributions are welcome though if you are interested in helping out.

from dask.

We've been adding type annotations continuously over the past 1-2 years but the code base is very large and ensuring that everything is typed is very difficult. This is rather an incremental process. As was already mentioned, contributions are welcome and I guess this is a comparatively easy topic for a first couple of contributions.

We're using mypy for type checking and the modules that are already fully typed are listed here (not many yet)

from dask.

@sam-goodwin (or anyone else having issues with type hinting) - are there any specific functions causing problems for you aside from array.rechunk? I'd be up for spending a little time working through some type annotations at some point - this issues currently labelled 'Add static type annotations' implying for the whole project, which is a great goal, but would be awesome to have a couple priority areas to target.

from dask.