move cloud fraction callback before rrtmgp

[szy21]

Purpose

This is at least one of the reasons that the all-sky radiation job is non-deterministic. Part of #2721. Also increases jet nfailures.

To-do

Content

---

• I have read and checked the items on the review checklist.

[szy21]

@charleskawczynski I think this should fix the non-deterministic issue, and I checked locally that the job is reproducible by looking at the norm of the solution. But is there an easy way to make sure the regression test works before merging it?

Edit: Actually it is still non-deterministic in some cases, which I think might be related to an issue in RRTMGP. I won't add back the regression test in this PR, but we should still have cloud fraction callback before rrtmgp I think.

[charleskawczynski]

Good catch! I'm not sure why this would result in new JET failures, though

[trontrytel]

Thank you for catching that!

[charleskawczynski]

Feel free to increase the number of allowed JET failures, so that we can merge this @szy21!

[szy21]

Sounds good, thanks. I will rebase and merge it now.