[fix](regression) Wait for stable colocate groups in plan tests

[morrySnow]

What problem does this PR solve?

Issue Number: None

Related PR: None

Problem Summary:

Colocate join plan assertions may run before ColocateTableCheckerAndBalancer marks the colocate group stable. In that state, Nereids can omit COLOCATE from the plan and make otherwise correct regression cases fail intermittently.

This PR adds a shared waitForColocateGroupStable helper to the regression test framework. It polls SHOW PROC '/colocation_group' until the target group's IsStable value is true, and fails on timeout. All positive COLOCATE plan assertions backed by explicit colocate_with groups now wait for stability first. The existing backup/restore-local waiting closure is replaced with the shared helper.

Release note

None

Check List (For Author)

• Test: Regression test
  • ./run-regression-test.sh --clean --run -d correctness_p0 -s test_colocate_join_of_column_order
  • ./run-regression-test.sh --run -d query_p0/join -s colocate_join_with_rollup
• Behavior changed: No
• Does this need documentation: No

[hello-stephen]

Thank you for your contribution to Apache Doris.
Don't know what should be done next? See How to process your PR.

Please clearly describe your PR:

1. What problem was fixed (it's best to include specific error reporting information). How it was fixed.
2. Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
3. What features were added. Why was this function added?
4. Which code was refactored and why was this part of the code refactored?
5. Which functions were optimized and what is the difference before and after the optimization?

[morrySnow]

/review

[morrySnow]

run buildall

[github-actions]

I found two issues to address before merge.

Critical checkpoint conclusions:

• Goal/test proof: The helper and most call sites address unstable colocate groups, but test_colocate_join_of_column_order still lets the negative plan assertions pass while the group may be unstable.
• Scope/focus: The change is small and focused.
• Concurrency/lifecycle/config/compatibility/transaction/persistence/data writes/observability: Not applicable; this PR only changes regression Groovy helpers/tests.
• Parallel paths: Positive COLOCATE assertions in the touched suites are covered; the first negative column-order path should use the same stable precondition.
• Test coverage/results: No .out files changed. I did not run regression tests in this review; I inspected the diff and ran git diff --check.
• User focus: No additional user-provided review focus was supplied.

[github-actions]

This shared helper does not handle Doris global colocate groups. colocate_with accepts names starting with __global__, and SHOW PROC '/colocation_group' reports those as the bare global name rather than db.__global__... (see ColocateTableIndex.GroupId.getFullGroupName). With the current unconditional prefixing, waitForColocateGroupStable("__global__...") will time out even when the group is stable. Please either leave global group names unprefixed here or make this helper explicitly db-scoped.

[github-actions]

This wait is after the four negative notContains "COLOCATE" assertions above, so those checks can still pass for the same unstable-group reason this PR is fixing. When the group is unstable, Nereids skips colocate join regardless of the join condition, which means these negative cases do not actually verify the column-order logic. Please move this wait to before the first explain after the inserts/planner setup so both the negative and positive assertions run with a stable colocate group.