Napari-feature-classifier

What's Changed
* Fix prediction/annotation layer issues by jluethi in https://github.com/fractal-napari-plugins-collection/napari-feature-classifier/pull/55


**Full Changelog**: https://github.com/fractal-napari-plugins-collection/napari-feature-classifier/compare/v0.2.0...v0.2.1

What's Changed
* Add napari 0.5.0 support & update prediction layer display by jluethi in https://github.com/fractal-napari-plugins-collection/napari-feature-classifier/pull/49


**Full Changelog**: https://github.com/fractal-napari-plugins-collection/napari-feature-classifier/compare/v0.1.2...v0.2.0

This release includes:
1. Handles translation in the label layers correctly (also during annotation)
2. Improve behavior upon Load Classifier: It shouldn't be required anymore to restart napari in regular workflows
3. Right-click an annotation to remove it
4. Pin napari to before 0.4.19 (because the label layer color updates break otherwise, see https://github.com/fractal-napari-plugins-collection/napari-feature-classifier/pull/40 for work to update it)

- Adds a new scikit-learn >= 1.2.2 dependency to solve https://github.com/fractal-napari-plugins-collection/napari-feature-classifier/issues/27
- Improve & test code path when features come without a roi_id column https://github.com/fractal-napari-plugins-collection/napari-feature-classifier/issues/28

Major refactor of the classifier:
- New & cleaner implementation of the whole workflow
- Classifier has improved performance of showing annotations
- Classifier now reads features from layer.features:
- Compatible with napari workflow outputs
- Still supports features from CSV files via a CSV loader
- Compatible with OME-Zarr-based images & measurements (using ROI loading plugin)

Fixing dependency issues that lead to napari crashing upon initialization of the annotator (see https://github.com/fractal-napari-plugins-collection/napari-feature-classifier/issues/24 )