Add IT for DELETE TIMESERIES replica consistency under IoTConsensusV2

[Pengzna]

Summary

• Add testDeleteTimeSeriesReplicaConsistency() integration test to verify that DELETE TIMESERIES operations are properly replicated across all DataNode replicas in a 3C3D IoTConsensusV2 cluster
• The test reproduces the scenario from the historical deletion replication bug (fixed in [IoTV2]: Pick deletion event for historical resend #17329): deletion events missing replicateIndex were silently dropped by IoTConsensusV2Processor, causing schema inconsistency across replicas
• Unify INSERTION constants across all test methods to use 3 columns (speed, temperature, power), replacing the prior 2-column variants

Test Scenario

1. Insert data with 3 measurements (speed, temperature, power) and flush
2. Insert more data without flush (WAL-only entries, simulating in-flight data)
3. DELETE TIMESERIES root.sg.d1.speed
4. Flush to persist the deletion
5. Wait for replication to complete (syncLag == 0) on all data region leaders
6. Verify schema consistency on each DataNode independently via SHOW TIMESERIES
7. Stop each DataNode one by one (triggers consensus pipe reconstruction + historical replay), verify all surviving nodes still show consistent schema — the deleted timeseries must be absent everywhere

Test Plan

• testDeleteTimeSeriesReplicaConsistency added in both stream and batch mode subclasses
• Run IoTDBIoTConsensusV2Stream3C3DBasicIT.testDeleteTimeSeriesReplicaConsistency
• Run IoTDBIoTConsensusV2Batch3C3DBasicIT.testDeleteTimeSeriesReplicaConsistency
• Verify existing tests test3C3DWriteFlushAndQuery and testReplicaConsistencyAfterLeaderStop still pass with unified 3-column insertions

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Adds an integration test to ensure DELETE TIMESERIES is consistently replicated across replicas in a 3C3D IoTConsensusV2 cluster (stream + batch), targeting a historical deletion-replication inconsistency. Also standardizes test insertions to use 3 measurements (speed, temperature, power) and updates existing verification accordingly.

Changes:

• Add testDeleteTimeSeriesReplicaConsistency() coverage in both stream and batch IoTConsensusV2 3C3D ITs via the shared base implementation.
• Implement end-to-end replica schema verification after DELETE TIMESERIES (including node stop/restart cycles) in the 3C3D base test class.
• Unify insert/query constants and expected verification results to use 3 measurements instead of prior 2-measurement variants.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
integration-test/src/test/java/org/apache/iotdb/db/it/iotconsensusv2/stream/IoTDBIoTConsensusV2Stream3C3DBasicIT.java	Exposes the new delete-timeseries replica consistency test in stream mode.
integration-test/src/test/java/org/apache/iotdb/db/it/iotconsensusv2/batch/IoTDBIoTConsensusV2Batch3C3DBasicIT.java	Exposes the new delete-timeseries replica consistency test in batch mode.
integration-test/src/test/java/org/apache/iotdb/db/it/iotconsensusv2/IoTDBIoTConsensusV23C3DBasicITBase.java	Adds the shared delete-timeseries replication test logic and updates inserts/verification for 3 measurements.

---

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

[Copilot]

After calling stopForcibly(), the test immediately asserts !isAlive(). AbstractNodeWrapper.stopForcibly() only waits up to 10s and ignores the return value, so the process may still be alive and this assertion can be flaky. Prefer awaiting the node to actually stop (e.g., Awaitility.until(() -> !stoppedNode.isAlive())) before proceeding.

Suggested change

	Assert.assertFalse(stoppedDesc + " should be stopped", stoppedNode.isAlive());
	Awaitility.await()
	.atMost(60, TimeUnit.SECONDS)
	.untilAsserted(
	() ->
	Assert.assertFalse(
	stoppedDesc + " should be stopped", stoppedNode.isAlive()));

[Copilot]

This log message line is likely to exceed the project's 100-character Checkstyle limit once indentation is included. Please wrap/split the string (or use multiple LOGGER.info calls) to keep each line within the limit.

Suggested change

	"Step 7: Stopping each DataNode in turn and verifying remaining nodes show consistent schema...");
	"Step 7: Stopping each DataNode in turn and verifying remaining nodes "
	+ "show consistent schema...");