For the sake of water tracers: Consider adding mixing parameterizations for lakes and possibly elsewhere

Emerging from [#518 ](https://github.com/escomp/ctsm/issues/518) (which was converted to [this identical discussion](https://github.com/ESCOMP/CTSM/discussions/2697)): I realized that there could be potential issues with the fact that for lakes – and possibly other things? – I believe that we currently don't compute any water fluxes between layers. I guess this is okay in the absence of water tracers, since the water volume is constant over time for each layer. But this seems more scientifically problematic with water tracers: I think the lack of fluxes between layers is equivalent to either assuming zero mixing between layers or instantaneous, perfect mixing between layers (off-hand, I'm not sure which).

So we may need to consider some kind of mixing parameterization and the calculation of explicit water fluxes between layers if we want to handle tracer concentrations rigorously.

We could also ignore this initially and come back to it later.

@nusbaume - do you know how this was handled in CESM1/CLM4? If not, we can dig into that. If it was ignored in CESM1/CLM4, maybe we can also ignore it for the initial implementation here?