Store object_id with links if available

[oruebel]

Fix #54

This PR updates the storage of links/references to add the following information:

• object_id: Object id of the reference object. May be None in case the referenced object does not have and assigned object_id (e.g., in the case we reference a dataset with a fixed name but without and assigned data_type (or neurodata_type in the case of NWB).
• source_object_id: Object id of the source Zarr file indicated by the source key. The source should always have an object_id (at least if the source file is a valid HDMF formatted file).

TODO:

• Updated the ZarrReference class to add a source_object_id and object_id keys
• Updated ZARRIO.__get_ref to populate the source_object_id and object_id keys
• Updated the storage documentation to document the source_object_id and object_id keys and update examples
• Update CHANGELOG
• Update tests

[codecov-commenter]

Codecov Report

All modified lines are covered by tests ✅

Comparison is base (e31f6a3) 85.66% compared to head (84240fb) 85.76%.

Additional details and impacted files

@@            Coverage Diff             @@
##              dev      #57      +/-   ##
==========================================
+ Coverage   85.66%   85.76%   +0.09%     
==========================================
  Files          13       13              
  Lines        3139     3189      +50     
==========================================
+ Hits         2689     2735      +46     
- Misses        450      454       +4     

Files	Coverage Δ
src/hdmf_zarr/backend.py	90.55% <100.00%> (+0.14%)	⬆️
src/hdmf_zarr/utils.py	96.49% <100.00%> (+0.69%)	⬆️
tests/unit/base_tests_zarrio.py	98.54% <100.00%> (+0.04%)	⬆️

... and 1 file with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[oruebel]

@mavaylon1 we should check whether this PR now also works with the fixes in #120 It would be nice we could include this in the release as well if it works. Otherwise, its fine to move this PR to the next release, but would be nice to push this over the finish line.

[oruebel]

Aside from adding/updating unit tests to check that the values for the object_id and source_object_id fields are correct, this PR should be ready.

[oruebel]

adding/updating unit tests to check that the values for the object_id and source_object_id fields are correct

Done

[mavaylon1]

@oruebel I wanted to as about the case when the object_id is None. You said that would be the case when the data_type/neuro_datatype is not assigned for a dataset. I was thinking of an example of what that would be. Say we have TimeSeries, which is a dataset. This has a type, but it also contains a dataset "data" that does not. This would be an example where we have "object_id" as none if that was the target correct?

[oruebel]

. This has a type, but it also contains a dataset "data" that does not. This would be an example where we have "object_id" as none if that was the target correct?

Correct. For TimeSeries.data the object_id would be None because it is just a dataset within a type. However, the source_object_id should always be present since the file is always represented by a Container. The object_id and source_object_id are not really being used right now, but will be useful to validate links and possibly in the future to be able to retrieve external links dynamically.