Update transform doc with random offset of padding due to pad_if_needed

[ekagra-ranjan]

In reference to #562 and #564 updating transform docs with the random offset of padding when pad_if_needed is set to True.

[codecov-io]

Codecov Report

Merging #791 into master will not change coverage.
The diff coverage is n/a.

@@          Coverage Diff           @@
##           master    #791   +/-   ##
======================================
  Coverage    38.1%   38.1%           
======================================
  Files          32      32           
  Lines        3128    3128           
  Branches      488     488           
======================================
  Hits         1192    1192           
  Misses       1857    1857           
  Partials       79      79

Impacted Files	Coverage Δ
torchvision/transforms/transforms.py	83.41% <ø> (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 138b5c2...ef2672c. Read the comment docs.

[fmassa]

The padding itself is not random:

vision/torchvision/transforms/transforms.py

Lines 464 to 468 in 4341640

	if self.pad_if_needed and img.size[0] < self.size[1]:
	img = F.pad(img, (self.size[1] - img.size[0], 0), self.fill, self.padding_mode)
	# pad the height if needed
	if self.pad_if_needed and img.size[1] < self.size[0]:
	img = F.pad(img, (0, self.size[0] - img.size[1]), self.fill, self.padding_mode)

but because we perform random crops, the padded area might be present in a seemly random position.

Maybe there is a way of clarifying this in the docs?

[fmassa]

Thanks!