Add notebooks for 001712

[alessandratrapani]

This PR adds an example notebooks for the 001712 dataset.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[bendichter]

Tested the merged branch and hit a data-side problem that I think needs to be resolved before this PR can run end-to-end. Surfacing here so you can decide how to proceed.

What's missing

All three notebooks expect data that isn't present in the published dandiset.

Inventory of 001712/draft: only 2 subjects, each with one desc-raw NWB file:

Path	Size
sub-FD-24/sub-FD-24_ses-..._desc-raw_behavior+ophys.nwb	(single raw file)
sub-FD-28/sub-FD-28_ses-81f90b18-..._desc-raw_behavior+ophys.nwb	33 GB

Notably absent:

• No subject CSK-im-011 (used as default in processed_widefield.ipynb).
• No desc-processed (or any other description besides raw) for any subject.

What the notebooks expect

anatomical_localization_widefield.ipynb does:

atlas_registration = nwbfile.lab_meta_data["atlas_registration"]   # KeyError
# … later …
localization = nwbfile.lab_meta_data["localization"]                # also missing
aci = localization.anatomical_coordinates_images["AnatomicalCoordinatesImageIBLBregma"]

I streamed the FD-28 file and inspected its actual structure: there's no lab_meta_data group at all. Top-level groups under /general/ are:

devices, ibl_metadata, institution, lab, optophysiology, protocol,
session_id, source_script, subject

Interestingly the ndx-anatomical-localization namespace is registered in the file (at /specifications/ndx-anatomical-localization/0.1.0/), but no instance of any of its types actually exists in the data. So the schema's there but never got populated.

Possible explanations

1. The notebooks were written against a dev version of the data that has these objects, and what's on DANDI is an earlier raw-only export.
2. The atlas-registration/localization step was supposed to be a separate upload (maybe a desc-processed file), which hasn't happened yet.
3. Two separate dandisets were merged conceptually and only one half made it to publication.

What I'd suggest

You probably know better than me which of the above is right. Two paths forward:

• Re-upload data with the atlas-localization NDX instances populated (and the CSK-im-011 / desc-processed files if those were also intended). The notebooks should then work as-is.
• Trim the notebooks to only demonstrate what's actually in the published files (subject/session loading, behavior + ophys access, etc.) and drop the atlas-localization sections — open a follow-up PR with the upload + re-add the atlas notebooks then.

Either way, this isn't a notebook-content bug I can patch — the data the notebook reads doesn't exist in the published dandiset yet. Let me know which direction you'd like to go and I can help.

For reference, the previous commits I pushed onto this branch (Colab bootstrap, dandi_authenticate removal, USE_DANDI default flip, relative-import fix) are still valid — they get the notebooks past the install + load stages before they hit this data-shape issue.