Different advection schemes for velocities and tracers

[glwagner]

We should allow different advection schemes for velocities and tracers. This requires doing something similar to what we do for closures. Perhaps we can allow

model = IncompressibleModel(advection=(velocities=CenteredFourthOrder(), tracers=WENO5()))

We also probably want to allow different advection schemes for every tracer. We can support something like

model = IncompressibleModel(advection=(velocities=CenteredFourthOrder(), T=WENO5(), S=CenteredSecondOrder()))

Let's discuss whether the syntax proposed above is acceptable or if we want a different design.

PS we need to write down a list of forbidden_tracer_names and throw an error if they are provided. Right now these are :u, :v, :w and any possible property of diffusivities. The above design would also preclude :velocities and :tracers.

[ali-ramadhan]

I think this syntax makes a lot of sense since it's pretty close to how users can specify different turbulent diffusivity closures for each tracer.

I guess you can't have different advection schemes for u, v, w so e.g. velocities=CenteredFourthOrder() makes sense too.

[francispoulin]

The setup @glwagner suggests for splitting the advection schemes would work well for IncompressibleModel, I think.

For ShallowWater, something different would need to be done. The solution field has uh, vh, and h, but it would be nice to have different advection schemes for the mass transports uh, and vh than for the mass equation h.

[glwagner]

Superceded by #2454