Three component phase separation for ultramassive white dwarfs

[mcastrotapia]

I added the files used in Castro-Tapia & Cumming (2025) to implement 3-species fractionation in white dwarf models. This includes the modified version of the phase_separation.f90 file and a directory in star/test_suite called wd_o_ne_3_phase to test the extension. Additionally, among the extensions of the phase_separation.f90, we added the isotope 22Ne to the previous O/Ne fractionation and fixed an inconsistency with the coordinate in mass of the solid core (see Appendix A of Castro-Tapia & Cumming, 2025).

[evbauer]

Thank you Matias! I'll need some time to review this code to make sure that I understand how it fits into MESA phase separation generally, but I'm happy to take a look and try to help resolve any issues that might come up.

We'll also want to mark all of the data files as git-lfs so that the git repository doesn't have to track very large diffs for those. https://docs.github.com/en/repositories/working-with-files/managing-large-files/configuring-git-large-file-storage

If you can push a commit to tag all of the .dat and .mod files here as git-lfs files, then I can change the target branch of the PR and go ahead and "squash and merge" this PR onto a testing branch here. Then we can continue to iterate by opening a new pull request onto that branch for any subsequent changes.

That may sound a bit convoluted, but the squashing + lfs files should allow us to avoid ever having very large diffs in the repository associated with this work, which I think will be worthwhile in the long run.

[mcastrotapia]

Hi Evan! Thank you for reviewing the pull request. I resubmitted the directory in the test_suite and marked all of the data files as git-lfs, so you can proceed. Please let me know if everything is okay :)

[evbauer]

Great! I ran the checks before squashing/merging, and it looks like the fortran linter caught some white space issues that would be good to clean up. I think it basically just wants to get rid of indendations on a bunch of blank lines. https://github.com/MESAHub/mesa/actions/runs/18109299040/job/51538440098?pr=860

The builds are also failing because the git checkout is failing to recognize the lfs files properly. I'm pretty sure the fix for that is just to add the .dat files to the top level .gitattributes files by adding this line alongside all the other LFS files listed there:

star/test_suite/wd_o_ne_3_phase/*.dat filter=lfs diff=lfs merge=lfs -text

Can you push changes to fix those two issues and we'll try again with the checks?

[mcastrotapia]

Hi Evan! I added the necessary files to .gitattributes. I hope it worked... I also eliminated the blank lines in the fortran files that were failing. Please let me know if the checks are okay now!

[evbauer]

It looks like we're also getting some relatively minor compilation failures: https://github.com/MESAHub/mesa/actions/runs/18114085086/job/51547218693?pr=860#step:3:8836

The fortran linter is picking up a few other things (trailing white space, lines that are technically too long, .eq. would be better as ==, etc), but we can fix those later. I'm happy to go ahead with the squash/merge once we get this passing the compilation tests, and then we can continue with a new PR from the testing branch here.

[evbauer]

Ah, I see now that we just lost a trailing underscore from end subroutine update_model_ in the latest commits, so I think compilation should be fine as soon as we get that one character back in there.

[mcastrotapia]

Thanks, Evan! I added the underscore. Let's see if it works