Make tracer vert adv always explicit

[charleskawczynski]

Closes #2510. @dennisYatunin, I think you meant to include q_liq/q_ice to be non-passive tracers, as I believe they would play a role in the dynamics for non-equilibrium cases, so I left those where they are. I didn't see any other tracers that needed to be moved in update_implicit_equation_jacobian!, but maybe @dennisYatunin can double check my work.

[dennisYatunin]

Although q_liq and q_ice play a role in the dynamics, their contribution to the thermodynamic state is negligible because they are several orders of magnitude smaller than q_tot. So, we are ignoring their contribution to the pressure gradient, which means that they do not play a role in the creation of fast waves. This means that we can treat them explicitly.

The changes I'm currently seeing move the advection tendencies for all tracers, including q_tot. The advection of q_tot must remain in the implicit tendency, as it plays a role in the creation of fast waves.

Also, the Jacobian needs to be updated. All blocks that correspond to the advection of passive tracers (this currently includes q_liq, q_ice, q_rai, and q_sno) should be removed. For example, the available_tracer_names in the advection_blocks should be replaced with ρq_tot_if_available in the ImplicitEquationJacobian constructor.

[charleskawczynski]

Alright, @dennisYatunin, I've updated based on your comments. How does this look?

[dennisYatunin]

Thanks, @charleskawczynski! I've fixed the remaining issues in this PR, and everything is now working as expected. However, the water budget conservation check is failing for the 1-moment microphysics test in CI. This is unfortunate, but it makes sense that we've introduced some additional error by making the advection of rain and snow at grid-mean velocity explicit. @szy21 and @trontrytel, we should discuss this result next week. If this change is not necessary, we might want to avoid it.