Add capability for user-defined parameter sets, with an example (take 2)

[glwagner]

This PR re-opens #172.

This PR tweaks the internal API for defining parameter sets to support user-defined parameters. This is important, because it will allow users to invoke functions in Thermodynamics.jl without requiring the full suite of parameters in ThermodynamicsParameters. For example, if one is only interested in computing density as a function of temperature, pressure, and (total) specific humidity, then they only require the universal gas constant and the molar masses of air and water.

Next, this PR adds an example which illustrates such a parameter set, called ConstitutiveParameters. It culminates in this plot:

which uses sea level data derived from the JRA55 reanalysis (subsampled to make the dataset smaller), showing the primary dependence of density on temperature, secondary dependence on pressure, and teritiary dependence on specific humidity for the typical conditions of the Earth's atmosphere at sea level.

Currently, there are no examples, so this PR introduces the first one. I'd be happy to turn the example into something that appears in the documentation via Literate (the example is very close to that form already).

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 90.93%. Comparing base (46f51af) to head (808b6b6).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #210   +/-   ##
=======================================
  Coverage   90.93%   90.93%           
=======================================
  Files          10       10           
  Lines        1180     1180           
=======================================
  Hits         1073     1073           
  Misses        107      107           

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

[glwagner]

@navidcy @charleskawczynski @simone-silvestri this passes now.

[glwagner]

@charleskawczynski can you please advise on how to make this more readable. It's better if all the arguments appear on one line.

[navidcy]

You need to enclose it within

#! format: off

#! format: on

Or just a #! format: off at the beginning of the .jl file to make the formatter ignore it altogether?

https://domluna.github.io/JuliaFormatter.jl/dev/#Turn-off/on-formatting

[glwagner]

Ok that's helpful. But why does the formatter want to do this in the first place?

[charleskawczynski]

From my point of view, the formatter is mostly a glorified line-breaker. It's great 95% of the time, but every now and then it's nice to use custom formatting.

I don't think there's an optimal solution for this sort of thing. Maybe something that uses ChatGPT could be a few percent better.

[glwagner]

But when does it try to break a line?

[charleskawczynski]

I think it's determined here: https://github.com/CliMA/Thermodynamics.jl/blob/main/.dev/clima_formatter_options.jl#L3

[charleskawczynski]

Can we please rebase and squash this PR? I like having a tidy and linear history.

[glwagner]

Why?

[charleskawczynski]

Why?

It's good practice. I can do it. Just let me know when you're ready to merge, @glwagner.

[glwagner]

I disagree that it's good practice for us. The PR is ready.

[navidcy]

I added the examples in the Docs via Literate.
Feel free to revert if you don't think this is a good idea.