adding coverage, found a bug going back to v0.0.11 with ravel... shape mismatches are the worst.
0.0.12
just a bunch of linting changes
0.0.11
uses code inspired by scikit-learn to handle different shapes, moves all the logic to a central processing function, and in doing so performs a significant refactor of the core functions.
0.0.10
going with the ravel() structure for a bit, may not persist. I want to respect the same shape-defaults as scikit learn does.
0.0.9
0.0.8
0.0.7 introduced an inadvertent bug wherein functionality for vector-valued norm evaluations was taken out by returning `result[0]` instead of `result` in `mynorm`. This fixes the behavior and enforces it with a test.