Upload pre-trained weights for MobileNet and ResNeXt

[fmassa]

Also move weights from ShuffleNet to PyTorch bucket.

I also renamed shufflenetv2 to shufflenet_v2 to make it consistent with the other models. This is fine because there hasn't been a release with shufflenet_v2 yet.

[codecov-io]

Codecov Report

Merging #917 into master will decrease coverage by 0.08%.
The diff coverage is 76.47%.

@@            Coverage Diff             @@
##           master     #917      +/-   ##
==========================================
- Coverage    57.9%   57.81%   -0.09%     
==========================================
  Files          46       46              
  Lines        3696     3703       +7     
  Branches      561      562       +1     
==========================================
+ Hits         2140     2141       +1     
- Misses       1429     1434       +5     
- Partials      127      128       +1

Impacted Files	Coverage Δ
torchvision/models/resnet.py	87.17% <100%> (ø)	⬆️
torchvision/models/mobilenet.py	92.3% <66.66%> (-4.25%)	⬇️
torchvision/models/shufflenetv2.py	86.04% <83.33%> (ø)	⬆️
torchvision/transforms/transforms.py	81.25% <0%> (-0.65%)	⬇️
torchvision/datasets/imagenet.py	21.55% <0%> (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 b384c4e...30e03da. Read the comment docs.

[barrh]

@fmassa are there plans for torchvision v0.3?

[fmassa]

@barrh yes, today :-)