Cityscapes Dataset Extraction (Follow-up PR to #1066)

[RJT1990]

Incorporating @fmassa and @TheCodez feedback from #1066:

• Now using the extract_archive function from dataset.utils
• Now autoextracts the trainextra dataset if split='train_extra' is specified

[codecov-io]

Codecov Report

Merging #1068 into master will increase coverage by 0.01%.
The diff coverage is 16.66%.

@@            Coverage Diff             @@
##           master    #1068      +/-   ##
==========================================
+ Coverage   63.87%   63.88%   +0.01%     
==========================================
  Files          66       66              
  Lines        5295     5294       -1     
  Branches      797      798       +1     
==========================================
  Hits         3382     3382              
+ Misses       1683     1682       -1     
  Partials      230      230

Impacted Files	Coverage Δ
torchvision/datasets/cityscapes.py	20% <16.66%> (+0.24%)	⬆️

---

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 867e5b4...8c40a25. Read the comment docs.

[fmassa]

LGTM, but we should really have tests for Cityscapes (but this is a separate feature request altogether)

@TheCodez can you review this PR as well?

[TheCodez]

LGTM, thanks

[RJT1990]

@fmassa : sounds good - I was going to submit a CamVid dataset PR so will write some new segmentation tests in that branch? (will submit later today)

[fmassa]

@RJT1990 please send a separate PR for the tests, following the dummy datasets approach that we currently have in test_datasets.py. Thanks a lot!