[Numpy][Operator] 'where' Implementation in MXNet

[hgt312]

https://docs.scipy.org/doc/numpy/reference/generated/numpy.where.html?highlight=where#numpy.where

[hgt312]

@mxnet-label-bot add [Numpy]

[xidulu]

Two question:

1. Numpy supports the following usage:

>>> import numpy as np
>>> a = np.random.uniform(size=(4,1))
>>> np.where(a > 0.5)
(array([0, 1, 2]), array([0, 0, 0]))

in short: x and y are optional ( either both or neither of x and y should be given)

2. How did you handle the scalar case mentioned in the docstring?

https://github.com/apache/incubator-mxnet/pull/16829/files#diff-2867f3bd5e893c32c128044e32d821d6R7353-R7359

[hgt312]

@xidulu

1. According to NumPy's doc, use np.nonzero, we have it.

When only condition is provided, this function is a shorthand for np.asarray(condition).nonzero(). Using nonzero directly should be preferred, as it behaves correctly for subclasses. The rest of this documentation covers only the case where all three arguments are provided.

2. Does not support scalar now, just use np.array to wrap it.

[haojin2]

LGTM other than the readability issue.