fix: Fixed PyTorch param loading

[fg-mindee]

Following up on #422, this PR switches back to the common parameter loading mechanism.

Previously, some pytorch models' factory functions were throwing errors when pretrained=True. With this PR, it throws a warning and proceeds with param initialization when no checkpoint is available.

Any feedback is welcome!

[charlesmindee]

Thanks

[codecov]

Codecov Report

Merging #425 (4923d1d) into main (94e984c) will increase coverage by 0.00%.
The diff coverage is 42.85%.

@@           Coverage Diff           @@
##             main     #425   +/-   ##
=======================================
  Coverage   95.80%   95.81%           
=======================================
  Files          96       96           
  Lines        3937     3942    +5     
=======================================
+ Hits         3772     3777    +5     
  Misses        165      165           

Flag	Coverage Δ
unittests	95.81% <42.85%> (+<0.01%)	⬆️

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

Impacted Files	Coverage Δ
doctr/models/recognition/crnn/pytorch.py	98.03% <33.33%> (+0.01%)	⬆️
...s/detection/differentiable_binarization/pytorch.py	96.36% <50.00%> (+0.03%)	⬆️
doctr/models/detection/linknet/pytorch.py	96.93% <50.00%> (+0.03%)	⬆️
doctr/models/backbones/mobilenet/tensorflow.py	97.50% <0.00%> (+0.06%)	⬆️

---

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 94e984c...4923d1d. Read the comment docs.

[charlesmindee]

Thanks!