WNet training

[C-Achard]

This PR implements WNet training directly in napari

• Local testing
• Upload new WNet weights based on new best model
• Coverage
• Docs -> done in extras WandB/Colab + docs #48

---

• Rebase to fix conflicts

[codecov-commenter]

Codecov Report

Merging #47 (328ef81) into cy/training (7a4e31f) will increase coverage by 1.27%.
The diff coverage is 80.16%.

@@               Coverage Diff               @@
##           cy/training      #47      +/-   ##
===============================================
+ Coverage        80.94%   82.22%   +1.27%     
===============================================
  Files               45       46       +1     
  Lines             4923     5614     +691     
===============================================
+ Hits              3985     4616     +631     
- Misses             938      998      +60     

Files Changed	Coverage Δ
napari_cellseg3d/code_models/worker_inference.py	75.56% <0.00%> (-1.49%)	⬇️
napari_cellseg3d/code_plugins/plugin_convert.py	89.63% <ø> (ø)
napari_cellseg3d/code_plugins/plugin_crf.py	78.32% <ø> (ø)
napari_cellseg3d/code_plugins/plugin_base.py	63.78% <36.00%> (-2.58%)	⬇️
napari_cellseg3d/code_models/worker_training.py	73.42% <71.65%> (+16.80%)	⬆️
napari_cellseg3d/code_models/model_framework.py	82.43% <80.00%> (+0.40%)	⬆️
...ri_cellseg3d/code_plugins/plugin_model_training.py	73.81% <80.31%> (+8.53%)	⬆️
napari_cellseg3d/interface.py	81.16% <87.50%> (+1.23%)	⬆️
napari_cellseg3d/utils.py	91.96% <93.33%> (+0.02%)	⬆️
napari_cellseg3d/_tests/fixtures.py	96.00% <100.00%> (+14.18%)	⬆️
... and 9 more

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more