Improve parametrization of Ion struct

[neil-glikin]

Particularly the energy levels -- there is a lot of redundancy for species with hyperfine structure. This also leads to redundancy in specifying Einstein-A coefficients for transitions. E.g. in yb171.jl:

("S1/2f=0", "P1/2f=1") => (multipole = "E1", einsteinA = 1.155e8),
("S1/2f=1", "P1/2f=0") => (multipole = "E1", einsteinA = 1.155e8),
("S1/2f=1", "P1/2f=1") => (multipole = "E1", einsteinA = 1.155e8),

All of these transitions are possible and have the same Einstein-A coefficient from atomic structure, hence the redundancy of specifying them all individually. (Yes, this is my fault.)