Dycore paper updates in how we compute SHF and LHF

[trontrytel]

Updates following section 3.6.4 in the dycore paper.

Tests work for me locally when using the Thermodynamics v0.13 branch. This PR should be merged after we do the Thermodynamics release.

cc: @kmdeck - will it affect the land models?

[trontrytel]

Sorry, none of the tests will work until we have Thermodynamics v0.13, because I put it as a requirement. So feel free to take a look at it after we have the release

[kmdeck]

@trontrytel In simulations where land is forced by a prescribed atmosphere, land needs to compute its own fluxes. In principle that means we just call SF functions, but instead right now we do that and then adjust the SF fluxes as follows:
https://github.com/CliMA/ClimaLand.jl/blob/a746febe976c4dcab555c2afb8dbbe0483e169b4/src/shared_utilities/drivers.jl#L367-L372

This is actually more similar to the SF computation in this PR, so that is good. We had switched to this form because it was easier to debug/understand than the previous SF computation (e.g. SHF doesnt depend on E). But note that we also need to account for surface resistance in the fluxes, and SF does not handle that right now, so we do it "offline".

To couple the full land model to atmos, we need to adjust SF to account for resistance. We should also adjust our standalone land model to use the output of SF directly. This will make coupled runs easier to debug.