Use environmental density consistency in diagnostic EDMF

[tapios]

In various places in diagnostic EDMF, we assume the product of density and area fraction in the environment, ᶜρa⁰, is equal to the density ᶜρ. This effectively leads to double counting of updraft tendencies in the grid mean (because the domain decomposition assumption is violated).

Harmonize and use the environmental density ᶜρa⁰ consistently.

[szy21]

It turns out we don't need it in other places other than microphysics. The only place that ᶜρa⁰ is used is in mass flux, but there is no mass flux contribution from the environment in diagnostic EDMF because we assume the environment tracer is the same as the grid-mean.