feat: Added shadow augmentation for all backends

[fg-mindee]

Following up on #730, this PR introduces the following modifications:

• adds RandomShadow augmentation
• adds unittests
• adds documentation

The following snippet:

import matplotlib.pyplot as plt
from doctr.io.image import read_img_as_tensor
from doctr.transforms import RandomShadow

transform = RandomShadow((.2, .8))
img = read_img_as_tensor("path/to/your/image.jpg")
plt.imshow(transform(img).numpy()); plt/axis('off'); plt.show()

yields:

Briefly:

• this first creates a 4-point polygon with 1 side on the top edge of the image
• the polygon is rotated & expanded
• the polygon is filled with 1s
• we apply an intensity gradient (so that the shadow fades as we go further from the source)
• with TF or PyTorch we overlay this after applying gaussian blur (a sharp shadow is not very believable)

Apart from filling the polygons, everything is in pure numpy until we overlay with the tensor image!

Any feedback is welcome!

[charlesmindee]

Thanks for this tranform, only a small comment

[codecov]

Codecov Report

Merging #811 (9b0433f) into main (bf2d4ca) will decrease coverage by 0.05%.
The diff coverage is 95.65%.

@@            Coverage Diff             @@
##             main     #811      +/-   ##
==========================================
- Coverage   96.01%   95.95%   -0.06%     
==========================================
  Files         131      131              
  Lines        4941     5073     +132     
==========================================
+ Hits         4744     4868     +124     
- Misses        197      205       +8     

Flag	Coverage Δ
unittests	95.95% <95.65%> (-0.06%)	⬇️

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

Impacted Files	Coverage Δ
doctr/transforms/modules/tensorflow.py	92.90% <90.90%> (-0.16%)	⬇️
doctr/transforms/modules/pytorch.py	97.14% <91.66%> (-1.17%)	⬇️
doctr/transforms/functional/base.py	95.31% <96.00%> (+1.56%)	⬆️
doctr/transforms/functional/pytorch.py	100.00% <100.00%> (ø)
doctr/transforms/functional/tensorflow.py	100.00% <100.00%> (ø)
doctr/models/detection/linknet/base.py	88.79% <0.00%> (-5.19%)	⬇️
doctr/models/predictor/tensorflow.py	100.00% <0.00%> (ø)
doctr/models/detection/linknet/pytorch.py	98.86% <0.00%> (ø)
doctr/models/detection/linknet/tensorflow.py	98.66% <0.00%> (ø)
doctr/models/predictor/pytorch.py	97.22% <0.00%> (+0.07%)	⬆️
... and 2 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 bf2d4ca...9b0433f. Read the comment docs.

[fg-mindee]

Ready for review @charlesmindee @SiddhantBahuguna 😁

[charlesmindee]

Thanks!