sparse.as_coo does not work for N-by-1 np.matrix

[gongliyu]

When input a N-by-1 matrix of type np.matrix, sparse.as_coo throws an error. It correctly dispatchs to sparse.COO.from_numpy but can not handle it. Please see the following

In [1]: import scipy.sparse,sparse

In [2]: sp = scipy.sparse.random(10,1,0.5)

In [3]: mat = sp.argmax(1)

In [4]: sparse.as_coo(mat)
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-4-5492fc8fff45> in <module>()
----> 1 sparse.as_coo(mat)

C:\Users\Liyu\Anaconda3\lib\site-packages\sparse\coo\core.py in as_coo(x, shape, fill_value)
   1749
   1750     if isinstance(x, np.ndarray):
-> 1751         return COO.from_numpy(x, fill_value=fill_value)
   1752
   1753     if isinstance(x, scipy.sparse.spmatrix):

C:\Users\Liyu\Anaconda3\lib\site-packages\sparse\coo\core.py in from_numpy(cls, x, fill_value)
    308             data = np.array(x, ndmin=1)
    309         return cls(coords, data, shape=x.shape, has_duplicates=False,
--> 310                    sorted=True, fill_value=fill_value)
    311
    312     def todense(self):

C:\Users\Liyu\Anaconda3\lib\site-packages\sparse\coo\core.py in __init__(self, coords, data, shape, has_duplicates, sorted, cache, fill_value)
    223                 msg = ('The data length does not match the coordinates '
    224                        'given.\nlen(data) = {}, but {} coords specified.')
--> 225                 raise ValueError(msg.format(len(data), self.coords.shape[1]))
    226             if len(self.shape) != self.coords.shape[0]:
    227                 msg = ("Shape specified by `shape` doesn't match the "

ValueError: The data length does not match the coordinates given.
len(data) = 1, but 0 coords specified.

[hameerabbasi]

This is now fixed and tested in master.