Synchronize np.load(allow_pickle=True) with scipy.sparse.load_npz

[doumbouyarene]

Can't use scipy.sparse.load_npz since it's uses numpy.load which now requires allow_pickle to be explicitly set to True.

Reproducing code example:

import numpy as np
import scipy.sparse as sparse

user_indicator_features = sparse.load_npz(user_feat_file)

Error message:

bash-3.2$ python main.py --task predict
Using TensorFlow backend.
1564089376.935315

Traceback (most recent call last):
File "main.py", line 130, in
user_indicator_features = sparse.load_npz(user_feat_file)
File "/anaconda3/lib/python3.7/site-packages/scipy/sparse/_matrix_io.py", line 131, in load_npz
with np.load(file, **PICKLE_KWARGS) as loaded:
File "/Users/rdoumbouya/.local/lib/python3.7/site-packages/numpy/lib/npyio.py", line 451, in load
raise ValueError("Cannot load file containing pickled data "
ValueError: Cannot load file containing pickled data when allow_pickle=False

Numpy/Python version information:

Python 3.7.3
Numpy 1.16.4
Scipy 1.2.1

[pv]

sparse.load_npz has always set allow_pickle=False, since files saved by save_npz do not (or should not) contain pickled data. `allow_pickle=True` will not be turned on, for security reasons. Most likely, there is some issue in the process that creates the file and ends up putting object arrays in it, even though those should not end up there.