More detailed tests for atom definitions

[hartytp]

Writing the tests, I was surprised how many small typos I found in the atomic structure data. Just silly things like an exponent missing in a transition frequency. It feels like there is still some data in this package which no one has ever used for a calculation. We should probably aim for every property of every atom to be tested in some way (e.g. transition frequencies can be compared by checking saturation intensity calculations). These don't all need to be ultra-precise - even just ruling out gross mistakes is quite useful.