Put back Letterbox transform

[MateoLostanlen]

After reading some disscutions on the ultralytics repo, I had removed the Letterbox preprocessing to gain in simplicity since it was considered unnecessary. I tested it on an image and obtained a similar result. After noticing some strange results this week, I measured the model's performance on an entire dataset using engine and got a much lower result. Using the Letterbox transformation, we get the same results as those obtained with the ultralytics library. Therefore, we need to add it again

Sorry for this error

[codecov]

Codecov Report

Attention: Patch coverage is 94.59459% with 2 lines in your changes missing coverage. Please review.

Project coverage is 92.30%. Comparing base (4c7aebf) to head (ff0237d).

Files	Patch %	Lines
pyroengine/utils.py	90.90%	2 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop     #212      +/-   ##
===========================================
- Coverage    92.38%   92.30%   -0.08%     
===========================================
  Files            6        6              
  Lines          486      507      +21     
===========================================
+ Hits           449      468      +19     
- Misses          37       39       +2     

Flag	Coverage Δ
unittests	92.30% <94.59%> (-0.08%)	⬇️

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

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.