Added napari plugin for reviewing and correcting bounding box annotations

[GauravSRC]

Summary

Closes #9.

Adds a napari plugin to ethology so that automatically generated bounding
box annotations can be loaded into napari, manually reviewed and corrected,
and then exported back to a COCO JSON file.

Changes

New package – ethology/napari/

File	Purpose
_reader.py	napari reader plugin: detects COCO/VIA JSON files and converts the ethology xr.Dataset to a napari Shapes layer (one rectangle per bounding box, coloured by category_id).
_writer.py	napari writer plugin: converts the (corrected) Shapes layer back to an ethology dataset and calls save_bboxes.to_COCO_file. Only rectangle shapes are exported; non-rectangle shapes are silently skipped.
widgets.py	make_correction_widget – a magicgui-based container with two panels: Load annotations (file picker + format selector) and Save corrected annotations (output file picker).
napari.yaml	Plugin manifest registering the reader, writer, and widget.

pyproject.toml

• Added [project.entry-points."napari.manifest"] for plugin discovery.
• Added napari optional-dependency group (napari>=0.4.19, magicgui, qtpy).
• Added [tool.setuptools.package-data] so napari.yaml is included in the sdist.

Tests

• tests/test_unit/test_napari/test_reader.py – unit tests for format
  detection, napari_get_reader, _reader_function, and shape geometry
  correctness.
• tests/test_unit/test_napari/test_writer.py – unit tests for
  _shapes_to_dataset, write_shapes, and two full round-trip tests
  (annotation count + coordinate fidelity).
• tests/test_unit/test_datasets/test_split.py – placeholder for future
  dataset-splitting tests.

How to test manually

pip install -e ".[napari,dev]"
napari path/to/annotation_file.json   # opens Shapes layer automatically
# or: Plugins → Annotation Correction → load file → correct → save

Notes

• The reader and writer are pure-Python and tested without a display.
• The widget tests are not included (they require a running Qt application);
  those will be added in a follow-up using pytest-qt.
• test_split.py is a placeholder; the split implementation is tracked
  separately.

[codecov]

Codecov Report

❌ Patch coverage is 76.36364% with 39 lines in your changes missing coverage. Please review.
✅ Project coverage is 78.85%. Comparing base (33adaa5) to head (dfddb67).

Files with missing lines	Patch %	Lines
ethology/napari/widgets.py	0.00%	36 Missing ⚠️
ethology/napari/_reader.py	96.20%	3 Missing ⚠️

❗ There is a different number of reports uploaded between BASE (33adaa5) and HEAD (dfddb67). Click for more details.

HEAD has 5 uploads less than BASE

Flag	BASE (33adaa5)	HEAD (dfddb67)
	10	5

Additional details and impacted files

@@             Coverage Diff             @@
##             main     #144       +/-   ##
===========================================
- Coverage   99.45%   78.85%   -20.60%     
===========================================
  Files           8       11        +3     
  Lines         554      719      +165     
===========================================
+ Hits          551      567       +16     
- Misses          3      152      +149     

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

🚀 New features to boost your workflow:

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