test(processing): avoid flaky incremental index cursor ordering

[FrankChen021]

Summary

Make IncrementalIndexCursorFactoryTest.testSanity independent of result sequence order.

GroupingEngine.process returns the raw grouped sequence and does not apply the query's ORDER BY post-processing. When the incremental index is not sorted by dimensions, positional assertions on the returned rows are not valid.

Evidence

The original failure occurred in the JDK 25 unit-test shard:

• Failed CI job
• Failing case: testSanity[1: onheap, sortByDim: false]
• Failure: expected the row beginning with null, but received the row beginning with hi

Before this change, the focused test was run 10 times locally on JDK 25 and reproduced the same failure twice. This confirms that the failure is nondeterministic rather than a deterministic assertion failure.

Changes

Replace positional row assertions with order-independent assertions while retaining validation of both expected result rows.

Tests

• Focused parameterized class: 16 tests, 0 failures, 1 skipped
• Focused testSanity repeated 10 times on JDK 25 after the change: 10/10 passed

[Copilot]

Pull request overview

This PR updates IncrementalIndexCursorFactoryTest.testSanity to avoid flaky assertions caused by nondeterministic row ordering from GroupingEngine.process when the underlying incremental index is not sorted by dimensions. It keeps validating both expected grouped rows while making the assertions order-independent.

Changes:

• Replace positional results.get(i) assertions with order-independent anyMatch checks against expected ResultRow arrays.
• Add an inline comment explaining why result ordering is not guaranteed in this execution path.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.