Get beavis-ci working

[drphilmarshall]

Figure out how to unit/continuously test notebooks, preferably with Travis CI - this needs setting up sooner rather than later, so we can use it to check our tutorials as we go. Belongs to epic #15

[drphilmarshall]

Update: the shared datasets are pretty big (~Gb). I wondered whether travis caching would help, but apparently not - because the files still need to be downloaded from the place that travis stores them: https://docs.travis-ci.com/user/caching/ . I'll try having travis curl down the ci_hsc shared dataset (3.6Gb) just to see what happens.

[drphilmarshall]

Change of plan: rather than pursuing travis-ci for our continuous integration needs, I am going to adapt the beavis-ci system that I started in the commissioning team notebooks repo and then just this week got working in DESC DC2. This is a simple(ish) bash script that is run by a cron job, and uses jupyter nbconvert to run the notebooks and push the rendered notebooks to an orphan branch for display.

[drphilmarshall]

This should have been closed by the messages in #62...