feat: Added rectangular stride MobileNets

[fg-mindee]

This PR introduces the following modifications:

• adds MobileNet V3 counterparts with 3 rectangular strides
• replaces mobilenet backbones in CRNN

I checked locally and the training now yields 92%+ of exact match after a single epoch (vs. 20% before)

Any feedback is welcome!

[charlesmindee]

Thanks!

[codecov]

Codecov Report

Merging #483 (869ff20) into main (a3f70e6) will decrease coverage by 0.42%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #483      +/-   ##
==========================================
- Coverage   95.43%   95.01%   -0.43%     
==========================================
  Files          97      108      +11     
  Lines        4035     4175     +140     
==========================================
+ Hits         3851     3967     +116     
- Misses        184      208      +24     

Flag	Coverage Δ
unittests	95.01% <100.00%> (-0.43%)	⬇️

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

Impacted Files	Coverage Δ
doctr/models/backbones/mobilenet/pytorch.py	100.00% <100.00%> (ø)
doctr/models/backbones/mobilenet/tensorflow.py	97.64% <100.00%> (+0.14%)	⬆️
doctr/models/recognition/crnn/pytorch.py	98.90% <100.00%> (-0.02%)	⬇️
doctr/models/recognition/crnn/tensorflow.py	100.00% <100.00%> (ø)
doctr/models/detection/core.py	97.61% <0.00%> (-0.66%)	⬇️
doctr/models/__init__.py	100.00% <0.00%> (ø)
doctr/models/detection/__init__.py	100.00% <0.00%> (ø)
doctr/models/recognition/__init__.py	100.00% <0.00%> (ø)
doctr/models/predictor/pytorch.py	38.46% <0.00%> (ø)
doctr/models/detection/predictor/__init__.py	100.00% <0.00%> (ø)
... and 13 more

---

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 a3f70e6...869ff20. Read the comment docs.