Migrate from AICSImage to BioImage

[jni]

Closes #100

Supersedes #123

From #123:

• Initial plan
• Replace aicsimageio with bioio in imports and class references

New changes:

• Add default physical pixel sizes to BioImage obj
• typo: filw -> file
• Add physical pixel size to BioImage in create_data
• Add bioio plugin dependencies
• Use separate resolutionunit for tifffile write
• Use bioio-tifffile to read raw.tif
• Fix patch to patch bioio instead of aicsimageio

I ran into bioio-devs/bioio-bioformats#40 which
means bioio-bioformats is probably not a good thing to rely on for now --
many dask workflows will fail with bioio-bioformats-read images.

Currently, all tests are passing on my machine except test_crop_workflow,
which fails with a PyClesperanto_prototype error. I'm hoping that will not be
an issue in CI/win/linux machines, let's see.

[jni]

@pr4deepr wanna approve the workflow to see if there are any further failures here?

[jni]

Well, that was disappointing 🤣

[jni]

Ok looks eminently fixable 😅 👍

[pr4deepr]

Python 3.8 and 3.9 errors are possibly due to bioio not supporting 3.10 - 3.13

However, error with tests for 3.10 appear to be pydantic related?

[jni]

due to bioio not supporting 3.10 - 3.13

You mean only supporting, yes? And yes, that's what I meant by eminently fixable... But I got stuck with other work... 😅

https://github.com/bioio-devs/bioio/blob/d6e614a4280f489877aa1fd3dd5d4f797ac57e2b/pyproject.toml#L25-L28

[jni]

(and yes the test matrix will need to be updated to only test 3.10+, which is fine since 3.9 is EOL.)

[multimeric]

Remind me why we can't use 3.11+? I think it's reasonable to drop 3.8-3.9 if we have to, because they're both end-of-life

[jni]

I think no reason @multimeric just needs updating the matrix

[jni]

I was just focused on getting stuff to run on my laptop when doing this work 😅

[multimeric]

I'm not sure, Pradeep specifically called it out in d7bfefa

[jni]

that was still in aicsimageio land

[jni]

?

[multimeric]

Well if you think the migration to BioIO unblocks the versions, then feel free to update the CI matrix

[jni]

at any rate I'm doing the WEHI bioinformatics seminar next week (Tue Dec 2), maybe we can get together in person and hammer this out? CC @jamesyan-git

[pr4deepr]

I wanted 3.8 and 3.9 because of some software packages I used during analysis, specifically ilastik and cellprofiler apis. Having aicsimageio in the same environment made it complicated as well.

Going ahead, it's going to be harder to maintain, so let's drop support for older python versions and add support for 3.11 & 3.12. If we need compatibility, will use older versions of napari-lattice.

Will have to add this in docs..