Add promoting constructors for Dual number support

[dennisYatunin]

This PR adds a method to every user-facing ThermodynamicState constructor, in which all input arguments are promoted to the same number type. In addition, q_vap_saturation_generic is given a new method that promotes its input arguments.

With only three exceptions, every function from Thermodynamics.jl that gets called while computing the precomputed quantities and implicit tendencies of ClimaAtmos.jl only takes two input arguments: a parameter set and a ThermodynamicState. The exceptions are

• constructors of ThermodynamicStates, where some inputs can be prescribed by the conditions at t = 0 or z = 0 while others depend on the prognostic state
• total_specific_enthalpy(param_set, ts, e_tot), where ts and e_tot both depend on the prognostic state
• q_vap_saturation_generic(param_set, T, ρ, phase), where T is prescribed by the conditions at z = 0 while ρ depends on the prognostic state in the bottom element

So, this PR constitutes the minimum set of additions to Thermodynamics.jl that will enable us to use automatic differentiation in ClimaAtmos.jl, which involves replacing values in the prognostic state with numbers of type ForwardDiff.Dual.

This is a simpler alternative to #197, which I'm going to close for now. In the future, we may want to come back to the changes in that PR, since they are much more generalizable. That is, they will not require to continue adding new methods to every function that relies on promotion during automatic differentiation in other repositories. The downside to those changes is that they would allow developers to silently introduce Float32/Float64 type instabilities within internal functions, and we would probably want to add more unit tests to ensure that those are avoided.

[codecov]

Codecov Report

Attention: 26 lines in your changes are missing coverage. Please review.

Comparison is base (1dc23bb) 93.04% compared to head (eccb2ae) 90.98%.

Files	Patch %	Lines
src/states.jl	0.00%	25 Missing ⚠️
src/relations.jl	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #199      +/-   ##
==========================================
- Coverage   93.04%   90.98%   -2.06%     
==========================================
  Files          10       10              
  Lines        1150     1176      +26     
==========================================
  Hits         1070     1070              
- Misses         80      106      +26     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.