revisit CMA

[zingale]

Currently we do the "consistent multifluid advection" method by constraining the species fluxes to sum to the density flux after we have done all of the transverse flux differences.

But CMA was developed for a directionally-split method. It might be the case that we want to normalize the fluxes after each Riemann solve, before they are applied to the conservative states in the transverse direction, as well as for the final construction.

Alternately, we might want to just switch to normalizes the species interface states themselves.

[zingale]

Our approach to CMA is to normalize the species fluxes to match the density flux after the Riemann solve. This is algebraically equivalent to the CMA approach of normalizing the final Riemann mass fractions, since the flux is linear in the mass fractions.

But we are only doing this for the final conservative update. I think that we want to do this for all Riemann solves, so it is picked up by the transverse solves too.