Update CI in MOM_interface so that "check_standalone_mom_build_and_run_lightweight_examples" works on MOM_interface forks

[manishvenu]

The check_standalone_mom_build_and_run_lightweight_examples previously worked by checking out the CESM, cd to the MOM_interface subfolder and checking out the active branch/pr. This logic of checking out MOM_interface does not work if the CI is on a different fork because the CESM only checks out ESCOMP/MOM_interface. The changes made here checks out the correct MOM_interface, and copies it over to the CESM for testing.

1. Add a step to check out MOM_interface to a temp folder
2. Replace the previous MOM_interface checkout with cp command to copy MOM_interface into the CESM folder.
3. Move the submodule consistency check before the MOM_interface checkout stage

[mnlevy1981]

We recently added a check to this CI to make sure that git fleximod checks out the correct submodules -- does this PR still run that test? We don't need to maintain the exact mechanism used previously, I just want to make sure this test will fail if the .gitmodules file tells git fleximod to check out a different tag than is checked out with git clone --recurse-submodules

[manishvenu]

We recently added a check to this CI to make sure that git fleximod checks out the correct submodules -- does this PR still run that test? We don't need to maintain the exact mechanism used previously, I just want to make sure this test will fail if the .gitmodules file tells git fleximod to check out a different tag than is checked out with git clone --recurse-submodules

Hey Mike, it should. I talked to Alper and moved that submodule consistency check up one step. It still does the check right after the CESM checkout but before we copy the current MOM_interface over into the CESM, which should still work out with what we want the test. It's just supposed to check that the current CESM checkout is doing the correct thing right? Nothing to do with whatever development is happening in the PR/Push.

[alperaltuntas]

My understanding is that, with these changes, the recently added "Check submodule hash consistency" step verifies the consistency of the MOM6 version that comes with CESM. However, we should be ensuring that the version introduced by a PR or push is consistent. To achieve this, the "Check submodule hash consistency" step should run after the "Copy the current MOM_interface to the correct place in CESM" step. Do you agree @mnlevy1981, @manishvenu ?

[manishvenu]

My understanding is that, with these changes, the recently added "Check submodule hash consistency" step verifies the consistency of the MOM6 version that comes with CESM. However, we should be ensuring that the version introduced by a PR or push is consistent. To achieve this, the "Check submodule hash consistency" step should run after the "Copy the current MOM_interface to the correct place in CESM" step. Do you agree @mnlevy1981, @manishvenu ?

Hey @alperaltuntas,

Since, we checkout the CESM and that checks out ESCOMP/MOM_interface, the submodule hash consistency test passed PRs and Pulls on ESCOMP/MOM_interface, but on forks of MOM_interface like manishvenu/MOM_interface copying the current MOM_interface failed. Would it be a compromise to run the submodule hash consistency after the Pull/PR ONLY if we are on ESCOMP/MOM_interface?

I don't know how the consistency test (CESM/bin/git-fleximod test) works, but I thought it didn't matter if the PR is before or after? I thought the test had something to do with CESM .gitmodules and fleximod, which wouldn't be impacted by what code is in MOM_interface (other than the git information).