Add support to py3.7 and pytorch > 1.0

1. In python3.7 "async" is used by python as a keyword which will result in a syntax error when used as a function parameter. Pytorch cuda semantics change it non_blocking
`async = True -> non_blocking = True`
2. volatile was removed in pytorch 1.0
> IBD-master/util/feature_operation.py:175: UserWarning: volatile was removed and now has no effect. Use `with torch.no_grad():` instead.
input_var = V(input,volatile=True)
3. Connot convert cuda tensor directly to numpy array

> Traceback (most recent call last):
File "test.py", line 20, in
features, _ = fo.feature_extraction(model=model)
File "/data4/lmx/tmp/IBD-master/util/feature_operation.py", line 180, in feature_extraction
while np.isnan(output.data.max()):
File "/data4/lmx/anaconda3/lib/python3.7/site-packages/torch/tensor.py", line 450, in __array__
return self.numpy()
TypeError: can't convert CUDA tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.

Modify line 180 in feature_operation.py will fix this issue
`while np.isnan(output.cpu().data.max()):`