RFC: Depend on obspec_utils for ObstoreReader, ObjectStoreRegistry

[maxrjones]

I think that ObstoreReader and ObjectStoreRegistry are useful outside of virtualizarr (e.g., for earthaccess) and should therefore be a separate library. I started obspec-utils accordingly, in line with some past discussions.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 88.75%. Comparing base (c67dcc1) to head (c2fcb6c).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #844      +/-   ##
==========================================
- Coverage   89.22%   88.75%   -0.48%     
==========================================
  Files          34       34              
  Lines        1987     1912      -75     
==========================================
- Hits         1773     1697      -76     
- Misses        214      215       +1     

Files with missing lines	Coverage Δ
virtualizarr/manifests/store.py	88.97% <100.00%> (ø)
virtualizarr/parsers/dmrpp.py	83.33% <100.00%> (-0.10%)	⬇️
virtualizarr/parsers/fits.py	100.00% <100.00%> (ø)
virtualizarr/parsers/hdf/hdf.py	95.41% <100.00%> (ø)
virtualizarr/parsers/kerchunk/json.py	100.00% <100.00%> (ø)
virtualizarr/parsers/kerchunk/parquet.py	90.69% <100.00%> (ø)
virtualizarr/parsers/netcdf3.py	100.00% <100.00%> (ø)
virtualizarr/parsers/typing.py	100.00% <100.00%> (ø)
virtualizarr/parsers/zarr.py	97.50% <100.00%> (-1.25%)	⬇️
virtualizarr/registry.py	100.00% <100.00%> (ø)
... and 2 more

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[TomNicholas]

This is technically a breaking change to the API, because the type of he ObjectStoreRegistry has changed. (Note that it literally changes the Parser protocol definition too). This justifies a minor version bump at least...

[TomNicholas]

Seems good to me

[maxrjones]

This is technically a breaking change to the API, because the type of he ObjectStoreRegistry has changed. (Note that it literally changes the Parser protocol definition too). This justifies a minor version bump at least...

I'll think on this a bit more. I forgot that a reason for obspec-utils was not only to have the code available as a lighter dependency but also to change the typing from obstore-based to obspec-based, which would be good to test before obspec-utils is a dependency of virtualizarr