Plot methods testing

[danilobellini]

Some AudioLazy methods lacks automated testing. Coveralls says the least covered file is the lazy_filters.py, and most of the tests it's lacking is about its interface with Matplotlib.

It seems difficult to test a plot. But some tests might be done on the output figure to see at least whether:

• Whether the output result is indeed a figure
• Assert if the amount of axes is ok
• Axes data should match expected values (for different input parameters)
• A circle should be printed in the zplot

A much more difficult test would be matching the resulting image with an image analysis procedure to "see" (literally) whether the elements (and their displacement) are correct. Testing pixel-by-pixel wouldn't help if small details are changed (e.g. font face in different systems), so that shouldn't be done.

For now (and for this issue), at least the basic internal information itemized above should be tested.

[goodboy]

Huh. Not that there's been much activity on this but just after reading this issue it made me think of this perceptualdiff project I had glanced at a while back. Using that to test mpl plots is kind of an interesting idea. Mind you, it's kind of a strange and monstrous test dependency.