Add distances

[lizliz]

We'll want to have lots of easy options for comparing ECTS:

• Standard L_p distances should be a one line command from scipy.distance
• Version of distance where we find the minimum L_p distance between a pair of matrices where we can spin one of the orientations and get the cheapest alignment. Dan has a version of this from his matisse example but I bet there's a matrix multiplication way to do this way faster.

[lizliz]

Additional Idea from Dan:

I was thinking more about PCA alignment. The orientation of a PCA axis is arbitrary, so I was thinking in comparing two ECTs, for one to flip the PC1 axis, flip the PC2 axis, and flip both (so one of the ECTs would now be four) and to take the minimum difference of that as the distance

And I think this corresponds to just a flip of the ECT matrix along with the rotations/sliding idea. Should be easy to do without recomputing the ECT.