Fix consistency of IAPWS water EoS

[speth]

Changes proposed in this pull request

• Fix dimensionalization of internal energy in WaterSSTP (was off by a factor of $T$)
• Correct conversion of nondimensional IAPWS equation of state to always be based on converting to a mass basis using the original defined value of $R$, then convert to a molar using the nominal molecular weight. This is the approach recommended by the IAPWS95 documentation.

A number of regression test values have changed, generally by factors on the order of $18.015268/18.015\approx 1.0000148$ which is the ratio of the molecular weight for water corresponding to the IAPWS95 value for $R$ and Cantera's molecular weight for water.

If applicable, fill in the issue number this pull request is fixing

Closes #1315

If applicable, provide an example illustrating new features this pull request is introducing

Checklist

• The pull request includes a clear description of this code change
• Commit messages have short titles and reference relevant issues
• Build passes (scons build & scons test) and unit tests address code coverage
• Style & formatting of contributed code follows contributing guidelines
• The pull request is ready for review

[bryanwweber]

I have no idea what's causing the Windows 2019 failures, that wouldn't also be the same on Windows 2022. Setuptools did release a new version today that updates distutils, but that doesn't seem to be fixed by capping the version of setuptools in pyproject.toml. Anyhow, #1392 (although currently DO NOT MERGE) enables running everything from Windows 2022 by fixing SCons/scons#3664, and all the builds on Windows 2022 succeed, so 🤷

I was trying to test over in #1392 what happens when SCons is < 4.4.0 and you have the msvc_toolset_version variable present, but I'm getting a shell error or something, so the failures are unrelated.

Also, the example failure is due to a new warning from Matplotlib 3.6.0, and us setting the environment variable to treat warnings as errors. I fixed that in #991 by installing Matplotlib <3.6.0, but it may be worth restricting the environment variable to only treat Cantera-generated warnings as errors.

I'm also very curious what you're working on that this was a priority to fix 😄 Unless it's just a general "well, I see how to fix this, so might as well" 😂

[codecov]

Codecov Report

Merging #1394 (64225fa) into main (5e15f57) will decrease coverage by 0.05%.
The diff coverage is 74.32%.

@@            Coverage Diff             @@
##             main    #1394      +/-   ##
==========================================
- Coverage   71.08%   71.03%   -0.06%     
==========================================
  Files         363      363              
  Lines       51839    51855      +16     
  Branches    17354    17362       +8     
==========================================
- Hits        36849    36834      -15     
- Misses      12645    12676      +31     
  Partials     2345     2345              

Impacted Files	Coverage Δ
include/cantera/thermo/WaterPropsIAPWS.h	100.00% <ø> (ø)
src/thermo/WaterPropsIAPWS.cpp	35.39% <65.85%> (-5.68%)	⬇️
src/thermo/PDSS_Water.cpp	72.26% <76.19%> (-0.28%)	⬇️
src/thermo/WaterSSTP.cpp	80.37% <100.00%> (+0.09%)	⬆️

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[speth]

I'm also very curious what you're working on that this was a priority to fix 😄 Unless it's just a general "well, I see how to fix this, so might as well" 😂

This isn't impacting anything specific I'm working on. Mostly just trying to make a little progress on the pile of bugs that were uncovered by #1299. I thought this was low-hanging fruit after figuring out the first one (scaling by R instead of RT), and then realized there were still some inexplicable discrepancies which were trickier to hunt down.

[ischoegl]

@speth … thanks for taking on some of the bugs. The code changes make sense to me (caveat: I have not reviewed underlying equations and am relying on the thermodynamics-inspired tests that uncovered the bug in the first place); I only have minor comments.

[ischoegl]

I have no idea what's causing the Windows 2019 failures, that wouldn't also be the same on Windows 2022. […]

The windows 2022 runner works a little different, as it builds in a conda environment.

[ischoegl]

Also, the example failure is due to a new warning from Matplotlib 3.6.0, and us setting the environment variable to treat warnings as errors. I fixed that in #991 by installing Matplotlib <3.6.0, but it may be worth restricting the environment variable to only treat Cantera-generated warnings as errors.

One other alternative would be to create a pytest runner that handles the examples, so we get access to the warnings filters.

PS: neither of the issues are related to the bug fix in this PR …

[speth]

I'm going to pass on trying to fix the pre-existing CI failures as part of this PR.

I briefly tried adjusting the PYTHONWARNINGS environment variable, and couldn't figure out how to get it to actually catch warnings from a specific module. I think PYTHONWARNINGS=error:::cantera should have been the right thing, but it didn't do anything when I deliberately triggered a warning from Cantera.