Improving unit tests for network extraction functions

[ma-sadeghi]

Some of our methods are not properly being tested by our unit tests. I only skimmed through test_network_extraction.py, and it seems that we're not properly testing network extraction. From what I see, most of the time, we're only checking the shape of the output, whereas at the very least, I think we should check the values.

[ma-sadeghi]

A good way to test the output value(s) of a function is to compare its np.linalg.norm, which sort of uniquely encapsulates the image values.

[jgostick]

I'm going to move this to a 2.1 issue. It's obviously important, but I think that a good thorough sweep of the unit tests is needed and that will take time and delay the release of 2.0.

[jgostick]

I have added buncha tests for the extractions. They check for exact numerical outputs, like the right number of peaks, regions, etc.