Use jsonschema for ROI collection validation

[niksirbi]

This PR contains an alternative implementation of ValidROICollectionGeoJSON, compared to the one found in #773.

If we decide that we prefer this one, we should merge this PR into the branch of #773. I'm personally undecided.

The ValidROICollectionGeoJSON validator was refactored from a manual validation approach to a schema-based approach. The following components were added:

• utils.py - Helper functions for loading schemas and validating files
• schemas/roi_collection.json - JSON Schema for ROI collection GeoJSON files

This PR description summarises the trade-offs.

Approach taken in PR 773

The original implementation used four separate @path.validator methods, each reading the file and checking specific conditions:

1. _file_is_valid_json - parsed JSON and caught JSONDecodeError
2. _file_is_feature_collection - checked type and features keys
3. _features_have_valid_geometry - iterated features to check geometry types
4. _roi_type_matches_geometry - validated cross-field consistency

Approach in this PR (JSON Schema)

The refactored implementation uses a JSON schema file (schemas/roi_collection.json) validated via jsonschema.validate(), plus one custom validator for cross-field logic that cannot be expressed in JSON Schema.

Pros of Cons of the JSON Schema Approach

Pros:

• Declarative validation logic: the schema file clearly documents what structure is expected.
• Separation of concerns: changes to the expected structure only require editing the schema file.
• Consistency with the approach used in ethology. If we go with this approach, we could even move some of the json schema-related utils that ethology has to movement.
• Reusability: The utility functions can be reused for other JSON-based file formats in the future (assuming we'll have any).

Cons:

• jsonschema as a new required dependency
• Less control over error messages. The jsonschema library generates its own error messages, which are technically accurate but less user-friendly than custom messages we define.
• Cross-field validation still requires Python.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (96239de) to head (9fb4264).

Additional details and impacted files

@@               Coverage Diff                @@
##           load_save_rois      #780   +/-   ##
================================================
  Coverage          100.00%   100.00%           
================================================
  Files                  35        37    +2     
  Lines                2239      2236    -3     
================================================
- Hits                 2239      2236    -3     

☔ 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.