cleanup format

[giovp]

addressed some of the points described in #149

• split format version per element

  • except for images and labels that are grouped together, following ome-zarr format

• removed @type and replaced it with encoding-type

• added a test for for shapes format

[giovp]

@ivirshup decided to go for spatialdata-encoding-type because I though:

• it's a safer option before things settle at ngff side.
• my understanding is that by registering it in spatialdata, it means that without spatialdata that anndata in the zarr store cannot be read, is that correct?

[codecov]

Codecov Report

Merging #154 (0a6e09d) into main (38c661c) will increase coverage by 0.04%.
The diff coverage is 91.89%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #154      +/-   ##
==========================================
+ Coverage   86.82%   86.86%   +0.04%     
==========================================
  Files          23       23              
  Lines        3377     3389      +12     
==========================================
+ Hits         2932     2944      +12     
  Misses        445      445              

Impacted Files	Coverage Δ
spatialdata/_io/format.py	87.61% <89.28%> (+1.47%)	⬆️
spatialdata/_io/read.py	99.36% <100.00%> (ø)
spatialdata/_io/write.py	97.31% <100.00%> (ø)

[LucaMarconato]

LGTM

[ivirshup]

it means that without spatialdata that anndata in the zarr store cannot be read, is that correct?

Yeah, unless the encoding type was just anndata, or we added the registration in anndata.

---

Overall looks good to me.

[giovp]

it means that without spatialdata that anndata in the zarr store cannot be read, is that correct?

Yeah, unless the encoding type was just anndata, or we added the registration in anndata.

makes sense, maybe worth it once tables are part of the ngff spec