fix: Fixed the box cropping from RandomCrop

[fg-mindee]

This PR introduces the following modifications:

• fixed the box cropping part of RandomCrop: a previous PR fixed the cropping of the image, but there was some offset left to correct and some rescaling to do
• updated unittests to avoid missing this next time
• took the liberty of refactoring numpy part into a shared crop_boxes function to avoid duplication

Any feedback is welcome!

[codecov]

Codecov Report

Merging #772 (cadeb52) into main (df162db) will increase coverage by 0.02%.
The diff coverage is 95.83%.

@@            Coverage Diff             @@
##             main     #772      +/-   ##
==========================================
+ Coverage   95.92%   95.94%   +0.02%     
==========================================
  Files         129      130       +1     
  Lines        4857     4857              
==========================================
+ Hits         4659     4660       +1     
+ Misses        198      197       -1     

Flag	Coverage Δ
unittests	95.94% <95.83%> (+0.02%)	⬆️

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

Impacted Files	Coverage Δ
doctr/transforms/functional/base.py	93.75% <93.75%> (ø)
doctr/transforms/functional/pytorch.py	100.00% <100.00%> (+2.27%)	⬆️
doctr/transforms/functional/tensorflow.py	97.72% <100.00%> (+1.57%)	⬆️
doctr/transforms/modules/base.py	94.52% <0.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update df162db...cadeb52. Read the comment docs.

[fg-mindee]

@charlesmindee the failing on test-evaluate is unrelated to this PR, so it's good for review 👍

[charlesmindee]

Thanks!