Low test coverage

[tsmbland]

Test coverage is pretty low, and should ideally be >80%. One problem is that a lot of the untested code is interactive matplotlib widgets. Not sure the best way of testing this code in an automated way, or whether it's better to leave these functions out of the coverage statistics? Or can just test that the figures open without error, without necessarily testing interactivity. Hmmm