Race condition in downloading model from model zoo in parallel

[eric-haibin-lin]

When i use horovod for training, and call

model = get_model(model_name, pretrained=True)

It complains with

Exception in thread Thread-5:
Traceback (most recent call last):
  File "/usr/lib/python3.6/threading.py", line 916, in _bootstrap_inner
    self.run()
  File "/usr/lib/python3.6/threading.py", line 864, in run
    self._target(*self._args, **self._kwargs)
  File "tests/python/unittest/test_gluon_model_zoo.py", line 32, in fn
    model = get_model(model_name, pretrained=True, root='parallel_model/')
  File "/home/ubuntu/src/mxnet/python/mxnet/gluon/model_zoo/vision/__init__.py", line 152, in get_model
    return models[name](**kwargs)
  File "/home/ubuntu/src/mxnet/python/mxnet/gluon/model_zoo/vision/mobilenet.py", line 375, in mobilenet_v2_0_25
    return get_mobilenet_v2(0.25, **kwargs)
  File "/home/ubuntu/src/mxnet/python/mxnet/gluon/model_zoo/vision/mobilenet.py", line 250, in get_mobilenet_v2
    get_model_file('mobilenetv2_%s' % version_suffix, root=root), ctx=ctx)
  File "/home/ubuntu/src/mxnet/python/mxnet/gluon/model_zoo/model_store.py", line 115, in get_model_file
    os.remove(zip_file_path)
FileNotFoundError: [Errno 2] No such file or directory: 'parallel_model/mobilenetv2_0.25-ae8f9392.zip'

The get_model API breaks if multiple processes are doing it at the same time.