Testing: use `np.array_equal()` instead of `(==).all()`

[dmdunla]

Suggestion: In testing, use np.array_equal() instead of (==).all() for better readability. Nick also uncovered a bug in the past using this approach that was not covered by all().

[ntjohnson1]

Re-found an example of the bug:

>>> import numpy as np
>>> ((2,2) == np.ones((2,2)) * 2).all()
True
>>> np.array_equal(np.ones((2,2))*2, (2,2))
False

[dmdunla]

Yes, I see that. Is this a result of numpy broadcasting in __eq__?

>>> import numpy as np
>>> (2,3) == np.ones((2,2)) * 2
array([[ True, False],
       [ True, False]])

[ntjohnson1]

I assume so. Either way array_equal is pretty obvious and clear. The example failure mode was just additional justification.