Non-Boussinesq vorticity

[johnomotani]

[Replaces #54 to use the branch on this repo]

Mostly copy-paste of bendudson/hermes-2#44.

Not sure about the multi-species aspects... Have just replaced Ne with N_sum and Pi with Pi_sum. Is this the right thing to do @bendudson?

Needs documentation.

Thanks @johnomotani ! Some things I'm unsure about:

• I think N_sum should be the mass density, so the sum of AA * N for all species which have a charge
• The average_atomic_mass setting needs thinking about: At the moment it defaults to 2, which is probably not the right choice. When the Boussinesq approximation is not used this setting should not affect the result, so possibly should not be read at all.
• The energy source term, which picks up a factor of N in non-Boussinesq case. I suspect this should be N_sum rather than N: If it's just N then the energy transferred to a species only depends on its temperature and is independent of its density. In the Boussinesq approximation the sum of AA * N is replaced with average_atomic_mass (total N normalised to 1) so I think this is consistent.