support for CICE6(or others) to rewind

[zhaobin74]

This PR closes #2959

Types of change(s)

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Trivial change (affects only documentation or cleanup)
• Refactor (no functional changes, no api changes)

Checklist

• Tested this change with a run of GEOSgcm
• Ran the Unit Tests (make tests)

Description

Allows a GC's refresh method (if present)to be called. Necessary for some GC's to rewind states.

Related Issue

[tclune]

This is ok by me, but I want @atrayano to review.

It is odd to use phase=MAPL_MAX_PHASES + 1, and am wondering if since this is not a run phase, it could use a different value.

[zhaobin74]

This is ok by me, but I want @atrayano to review.

It is odd to use phase=MAPL_MAX_PHASES + 1, and am wondering if since this is not a run phase, it could use a different value.

@tclune, I don't like hardcoded 1 either. But it is the only one I can come up with which works. Maybe @atrayano can fix it.

[mathomp4]

@zhaobin74 How crucial is this?

If you'd like a release but can wait until next week, we can look at getting out a new MAPL release. develop has been accumulating a lot.

But, if you need it urgently, I can make a hotfix and release a new MAPL 2.47

[zhaobin74]

@zhaobin74 How crucial is this?

If you'd like a release but can wait until next week, we can look at getting out a new MAPL release. develop has been accumulating a lot.

But, if you need it urgently, I can make a hotfix and release a new MAPL 2.47

@mathomp4 , it is not urgent. @Dooruk is on PTO and he can already run his DA tests using my build (v1160) with the MAPL branch I made these changes on. I think it should be more vigorously vetted and hopefully @atrayano could further refine it.

[atrayano]

@zhaobin74 I have several concerns regarding this PR. The most trivial one is that instead of hardcoding MAPL_MAX_PHASES+1, we could introduce a local variable "phase", which we should get by calling ESMG_GridCompGet(gc, currentPhase=phase, _RC). A far more serious concern is the fact that we are using cold_start method which was supposed to be only as a fallback strategy during Initialize. Luckily for us, thus method is not used anywhere in GEOS (with a possible exception of the dycore), but if another component legitimately tries to use it, this PR has the potential to break that imaginary component. Let's touch base on Monday to brainstorm and find a better solution.

For the time being, I will add "do not approve" label.

[tclune]

See inline comments.