Do we want to support tensors with boolean values?

[dmdunla]

This is a change from MATLAB TTB, where boolean tensors are represented using values of 1 (true) and 0 (false), rather that as true boolean tensors:

>> T = tensor([1,2,3,4],[2,2]);
>> T > T
ans is a tensor of size 2 x 2
	ans(:,:) = 
	   0   0
	   0   0

And here is the corresponding pyttb code:

>>> T = ttb.tensor(np.array([1,2,3,4]),(2,2))
>>> T > T
tensor of shape (2, 2)
data[:, :] =
[[False False]
 [False False]]

We currently do not explicitly manage dtypes of tensors in array (analogous to numpy), although we use numpy arrays for much of the underlying data in pyttb tensors, so we inherit some of that functionality.

If we want to support this, this would require revisiting constructors and methods that involve boolean values.

Suggestion: push this to post-2.0.