Add features from cellmap fork

[d-v-b]

@rhoadesScholar can you summarize the changes from the cellmap fork?

[rhoadesScholar]

changes were made by @yuriyzubov, so he'd probably give the best summary.

[mzouink]

@yuriyzubov can you please explain what you did ?

[rhoadesScholar]

These changes facilitate use of funlib.persistence with multi-resolution OME-NGFF Zarr containers, as used by CellMap. This is a core dependency to DaCapo and needs to be updated on PyPi before DaCapo can be published.

@yuriyzubov Are there any further additions you would like to make?

[yuriyzubov]

1. For _read_voxel_size_offset method:

   1. The store (path to the zarr/n5 container) and path (location of the array within zarr/n5 container) of the input zarr/n5 array is regularized with separate_store_path() method
   2. For zarr container, multiscale is not always located in the parent group metadata of the zarr array, so check_for_multiscale() traverses through all parent groups of the array until it finds a multiscale attribute.
   3. The information about scale is not complete without units, thus search for units is realized with check_for_units().

2. For zarr container, when calling prepare_ds() method, adding multiscale attribute to parent group metadata is realized through add_multiscales_metadata() method.

[mzouink]

@yuriyzubov @d-v-b Can you please take a look here
data :
/nrs/cellmap/data/jrc_mus-liver-zon-1/jrc_mus-liver-zon-1.n5/em/fibsem-uint8/
/nrs/cellmap/zouinkhim/data/tmp_data_v3/jrc_mus-liver-zon-1/jrc_mus-liver-zon-1.n5/volumes/groundtruth/crop345/labels/mito

After the PR the axes are flipped for GT in N5

[d-v-b]

I can confirm that /nrs/cellmap/zouinkhim/data/tmp_data_v3/jrc_mus-liver-zon-1/jrc_mus-liver-zon-1.n5/volumes/groundtruth/crop345/labels/mito is not aligned to the FIB-SEM data, but I need to see the code you used to generate that image in order to diagnose the issue.

[d-v-b]

@rhoadesScholar I unassigned myself, because it's not clear what you need from me in this PR

[rhoadesScholar]

@rhoadesScholar I unassigned myself, because it's not clear what you need from me in this PR

I assigned you, @d-v-b , because you created the PR and pair well with @yuriyzubov to complete remaining issues. After that I'll merge it. Thanks!

[d-v-b]

complete remaining issues.

maybe list these issues here? I don't know what they are :)

[d-v-b]

to be clear, I opened the PR because someone needs to open a PR to bring the cellmap stuff into this repo, but none of the commits in here are mine 🤣, and I'm not a maintainer of this repo

[mzouink]

The remaining issue:
Using the current version of funlib.persistence the N5 datasets aligns togethers but using the PR version they don't align anymore. Seams like default offset order for N5 is wrong