feat: enable sorting for CH with list variable

[zepfred]

No description provided.

[zepfred]

@triceo I noticed other factories used by configurations and annotations that are not located in the public API. Maybe, we should handle that in a separate PR as well.

[triceo]

As always, the different selector combinations are causing me to ask questions - see inline.

Downstream tests are failing.

Also, please provide a migration recipe for the move of the interface to the public API; the uses of the old interface should be replaced by the new one.

Finally, the change needs to show up in the upgrade recipe in our docs.

[zepfred]

Finally, the change needs to show up in the upgrade recipe in our docs.

I have included a recipe for automatic migration to the new interface. Do we need to add anything to our docs?

[triceo]

General comments:

• Deprecated declarations should move towards the end of the file. We want to see the active code first.
• Plenty of places use the word "sorterComparator" where it should just be "comparator"; other places already use "comparator". This needs to be standardized.
• If we ever touch code that uses @NonNull, let's convert it to @NullMarked. Reduces boilerplate.

Most importantly though - this PR has the same potential to break the solver, just like 1.26.0 did. What can we do to ensure that the same problem isn't repeated?

I suggest you ask Marek to try this with his sorting PoC in FSR. That will prove it still does what it needs to do. Carefully testing ESS will ensure the old code still works, because AFAIK ESS was already using it.

I am not yet done with the review, but this should be enough for now. With this, the codebase will change enough that I'll have to re-read it anyway.

[triceo]

Forgot to submit these comments as well.

[triceo]

Big PR! Finally got through all of it.

Leaving some more comments, but I think we're getting there.
None of these changes will affect performance of the code, therefore the existing experiments are not invalidated.

[triceo]

LGTM, assuming that the verification on models passes.

[triceo]

Anything you still want to do, @zepfred?
Since you're gone the rest of the week, I want to be sure before I merge it.

[sonarqubecloud]

Quality Gate passed

Issues
1 New issue
0 Accepted issues

Measures
0 Security Hotspots
88.9% Coverage on New Code
2.6% Duplication on New Code

See analysis details on SonarQube Cloud