Update BeamEmissionMultiplet with more accurate MSE calculation

[Balazs-Peter]

The spectrum calculation in the current BeamEmissionMultiplet class uses constant line intensity ratios, but in a real situation the ratios also depend on the observation direction. Better intensity ratios, or a method to calculate them might also be available.

[vsnever]

Hi @Balazs-Peter, sorry for not replying for so long.

There is a PR #400, which moves the line shape models to a dedicated submodule (see #396). I think it's better to wait for #400 to be merged before creating a PR for BeamEmissionMultiplet.

In terms of implementation, BeamEmissionMultiplet should not have atomic data hard-coded. At the moment, atomic data distributed within Cherab goes to cherab/core/atomic/data, and models access it through the AtomicData interface. Also, I would like BeamEmissionMultiplet to be compatible with Marchuk's data as the most advanced atomic data for MSE at the moment, even if we can't distribute it within Cherab. The latter means that all component ratios (σ₁/σ₀, π₂/π₃, etc.) must be functions of the beam energy and electron density.