Build regression tests and a harness

[kokes]

I'm not sure what the best course of action here is. Whether we should just care about DOM generation or if we should go full "screenshot testing" - using Selenium or Puppeteer.

The main reason is to test things like overflows in pres and imgs, multiple outputs, legacy versions etc. Things that tend to break with changes to the library code.

[kokes]

We could also leverage the nbformat and nbconvert tests

• https://github.com/jupyter/nbformat/tree/master/nbformat/tests
• https://github.com/jupyter/nbconvert/tree/master/nbconvert/tests/files

[kokes]

This commit 36c745d could be tested for regressions using https://github.com/jupyter/nbconvert/blob/master/nbconvert/tests/files/notebook1.ipynb