I now saw that @giovp used 1.9.1 as lower bound for conda-forge. We could therefore consider using 1.9.1 instead of 1.8.0 in this PR; or we could relax the conda-forege contraint. Giovanni WDYT?

I chose this version because from spatial-image==1.2.0 the lower bound xarray>=2024.10.0 is added, and because xarray-dataclasses==1.8.0 supports xarray>=2024.10.0 but xarray-dataclasses==1.7.0 does not.

The package manager should be able to resolve the dependency tree correctly on its own. However, there are issues in transitive package dependency specification as discussed in #28 and reported in astropenguin/xarray-dataclasses#240. Adding a further valid constraint to address some package manager's dependency resolution as workaround is ok, but it is does not resolve the underlying issue. In general, adding unnecessary constraints has the potential to cause issues with other packages in a package or project and should not be added unless needed.

I now saw that @giovp used 1.9.1 as lower bound for conda-forge. We could therefore consider using 1.9.1 instead of 1.8.0 in this PR; or we could relax the conda-forege contraint.

No, this additional constraint will cause more issues re: zarr 3 per astropenguin/xarray-dataclasses#240