Add testing for notebooks [WIP]

[kmader]

I started to implement some basic testing for the notebooks in response to the comment here: https://github.com/tensorflow/lucid/pull/97#issuecomment-411925091

[kmader]

The biggest open issues now are

• [ ] print statements in python 3
• [ ] !pip install messing up configurations

[ludwigschubert]

Oh, thanks, I’ll take a look! I had found https://github.com/computationalmodelling/nbval which uses magic comments to to mark cells to be skipped over. I was mostly stuck on how to get non-random outputs, which seems critical for equality comparison. Print statements etc we’ll just port to Python 3; we primarily want the main library to be backwards compatible for users stuck on Python 2 for now.

That being said, even just running the code would provide a layer of stability that lets us know when we break things. Thanks already for getting started; I may need some days to respond!