better test coverage

[rkurchin]

Coverage isn't terrible, but there are some particular corners that are currently left by the wayside...

Particularly (in roughly decreasing order of importance as I see it):

• phase_diagram for different model types, different parameter values, deintercalation case
• quantum capacitance functionality
• integrand function
• model fitting
• is there a good way to test plotting functions?

Leaving this here as a note for the future if I (or any contributors...?) want to address it...