Disable pretrained backbone downloading if pretrained is True

[vfdev-5]

Fixes #3322

Description:

• Avoid downloading backbone weights and then segmentation model weights:
  Actual behaviour:

python -c "import torchvision as tv; tv.models.segmentation.fcn_resnet101(pretrained=True)"
Downloading: "https://download.pytorch.org/models/resnet101-5d3b4d8f.pth" to /root/.cache/torch/hub/checkpoints/resnet101-5d3b4d8f.pth
100%|█████████| 170M/170M [00:01<00:00, 99.8MB/s]
Downloading: "https://download.pytorch.org/models/fcn_resnet101_coco-7ecb50ca.pth" to /root/.cache/torch/hub/checkpoints/fcn_resnet101_coco-7ecb50ca.pth
100%|█████████| 208M/208M [00:02<00:00, 102MB/s]

and new behaviour

python -c "import torchvision as tv; tv.models.segmentation.fcn_resnet101(pretrained=True)"
Downloading: "https://download.pytorch.org/models/fcn_resnet101_coco-7ecb50ca.pth" to /root/.cache/torch/hub/checkpoints/fcn_resnet101_coco-7ecb50ca.pth
100%|█████████| 208M/208M [00:02<00:00, 88.5MB/s]

[datumbox]

LGTM, thanks!

[codecov]

Codecov Report

Merging #3325 (eb16bd8) into master (6116812) will increase coverage by 0.11%.
The diff coverage is 0.00%.

@@            Coverage Diff             @@
##           master    #3325      +/-   ##
==========================================
+ Coverage   73.84%   73.96%   +0.11%     
==========================================
  Files         104      104              
  Lines        9606     9607       +1     
  Branches     1537     1537              
==========================================
+ Hits         7094     7106      +12     
+ Misses       2033     2024       -9     
+ Partials      479      477       -2     

Impacted Files	Coverage Δ
torchvision/models/segmentation/segmentation.py	74.07% <0.00%> (-0.93%)	⬇️
torchvision/datasets/video_utils.py	69.27% <0.00%> (ø)
torchvision/models/detection/_utils.py	82.58% <0.00%> (ø)
torchvision/models/quantization/resnet.py	93.25% <0.00%> (ø)
torchvision/models/quantization/googlenet.py	64.77% <0.00%> (ø)
torchvision/models/quantization/inception.py	30.89% <0.00%> (ø)
torchvision/models/detection/backbone_utils.py	88.05% <0.00%> (ø)
torchvision/models/quantization/shufflenetv2.py	90.16% <0.00%> (ø)
torchvision/transforms/functional_tensor.py	79.23% <0.00%> (+0.19%)	⬆️
torchvision/models/detection/anchor_utils.py	94.66% <0.00%> (+1.33%)	⬆️
... 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 6116812...c19d9fc. Read the comment docs.