[Test] Stabilize table DDL HA tests for async DROP TABLE

[Caideyipi]

Description

After #18806 made DROP TABLE asynchronous, IoTDBTableDDLHAIT still checked SHOW TABLES immediately after submission. Both the stopped-DataNode and read-only-DataNode cases can fail while the drop procedure is still running.

Wait up to 60 seconds for the latest matching procedure in information_schema.drop_table_procedures to reach SUCCESS, then retain the table-absence assertion before dropping the database. Retry transient inconsistencies between cluster progress snapshots while waiting, and include the procedure error message in the state assertion.

Validation

• Spotless formatting and checks passed.
• Integration test sources compiled successfully.
• git diff --check passed.
• IoTDBTableDDLHAIT passed locally in TableClusterIT mode on JDK 17: 2 tests, 0 failures, 0 errors, 0 skipped. Both the stopped-DataNode and read-only-DataNode scenarios passed; Failsafe verification reported BUILD SUCCESS.
• Rebuilt the cluster template and ran mvn -nsu failsafe:integration-test@integration-test failsafe:verify@verify -pl integration-test -am -PTableClusterIT -Pwith-integration-tests -Dit.test=IoTDBTableDDLHAIT -DfailIfNoTests=false -Dfailsafe.failIfNoSpecifiedTests=false with bounded JVM heaps.

This PR has:

• been self-reviewed.
• added comments explaining the asynchronous wait.
• modified the existing integration tests for both HA scenarios.
• been tested in a test IoTDB cluster.