Use trainer.predict for single images

[jveitchmichaelis]

Description

Use the trainer for single image prediction, similar to other predict routes (predict_file).

I've used SingleImage as the dataset which seems sensible, but this accepts a patch_size option which is not the behavior we had before. Prediction would use the entire image. Here I've forced patch_size to be the same size as the image only in deepforest.predict_image which is a hack, but the behaviour is the same as before.

I've refactored _dataloader_wrapper_ for convenience and added crop_model support for single images. It now uses _crop_models_wrapper_ which has the same logic for multiple models.

There are some inconsistencies that we could look into - predict_file returns an empty dataframe for no predictions, while predict_tile and predict_image (previously and here) return None. Some prediction paths convert labels from int->str - plan to make everything consistent there?

Related Issue(s)

Would supersede and fix #1391

AI-Assisted Development

N/A

[jveitchmichaelis]

Ready for review and we can address the inconsistency separately. For now this should fix the device placement bug.

[codecov]

Codecov Report

❌ Patch coverage is 75.75758% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 85.26%. Comparing base (d2851ec) to head (77881cf).
⚠️ Report is 9 commits behind head on main.

Files with missing lines	Patch %	Lines
src/deepforest/predict.py	63.15%	7 Missing ⚠️
src/deepforest/main.py	92.85%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1398      +/-   ##
==========================================
- Coverage   86.61%   85.26%   -1.36%     
==========================================
  Files          26       27       +1     
  Lines        3736     3875     +139     
==========================================
+ Hits         3236     3304      +68     
- Misses        500      571      +71     

Flag	Coverage Δ
unittests	85.26% <75.75%> (-1.36%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 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.