Use OME v0.5 by default and require it for Zarr v3

[kabilar]

Context

Currently it looks like the default Zarr version is 3 but the default OME version is 0.4. Changing this default uncovered some issues with linc-convert (e.g. golden test data) and nifti-zarr-py.

Changes

• Update the default OME version to 0.5 and require it for Zarr v3
• Pass ome_version to write_ome_metadata across all modalities
• Derive ome_version from zarr_version in tests and regenerate golden test data
• Depends on Fix OME v0.5 metadata neuroscales/nifti-zarr-py#51
• Depends on new nifti-zarr-py release

cc @satra @ayendiki

[kabilar]

The CI is failing. Something looks quite off with regards to the test data: https://github.com/lincbrain/linc-convert/tree/main/tests/data

The zarr v2 test data have zarr.json files. And the zarr v3 zarr.json files have the ome key set to 0.4.

[kabilar]

The CI is failing. Something looks quite off with regards to the test data: https://github.com/lincbrain/linc-convert/tree/main/tests/data

The zarr v2 test data have zarr.json files. And the zarr v3 zarr.json files have the ome key set to 0.4.

Hi @balbasty, as I am working to fix the test data, I am thinking that we should drop support in linc-convert for Zarr v2/OME v0.4. I don't think we have any data associated with publications or newly acquired data that is Zarr v2/OME v0.4. Although it is hard to be sure since the zarr.json files have the "version": "0.4", for example with the M1 LSM: https://webdav.dandiarchive.org/dandisets/001412/draft/derivatives/sub-MF283/strips_raw/sample-slice001/sub-MF283_sample-slice001_chunk-0001_acq-camera-01.ome.zarr/. What do you think?

[kabilar]

The CI is failing. Something looks quite off with regards to the test data: https://github.com/lincbrain/linc-convert/tree/main/tests/data

The zarr v2 test data have zarr.json files. And the zarr v3 zarr.json files have the ome key set to 0.4.

I think I have fixed these issues, but neuroscales/nifti-zarr-py#51 has to be merged and released first.

[kabilar]

With the new release of nifti-zarr-py, the CI now pass and this pull request is ready for review.

https://pypi.org/project/nifti-zarr/#history