[fix][test] Stabilize delayed auditor lost bookie tests in AuditorLedgerCheckerTest

[oneby-wang]

Motivation

AuditorLedgerCheckerTest.testDelayedAuditOfLostBookies is flaky when repeated with a high invocation count. The test configures lostBookieRecoveryDelay to 5 seconds, shuts down a non-auditor bookie, and then uses fixed waits that start immediately after the shutdown thread is launched.

audit of lost bookie isn't delayed
java.lang.AssertionError: audit of lost bookie isn't delayed
	at org.testng.AssertJUnit.fail(AssertJUnit.java:65)
	at org.testng.AssertJUnit.assertTrue(AssertJUnit.java:23)
	at org.apache.bookkeeper.replication.AuditorLedgerCheckerTest.testInnerDelayedAuditOfLostBookies(AuditorLedgerCheckerTest.java:415)
	at org.apache.bookkeeper.replication.AuditorLedgerCheckerTest.testDelayedAuditOfLostBookies(AuditorLedgerCheckerTest.java:436)
	at java.base/jdk.internal.reflect.DirectMethodHandleAccessor.invoke(DirectMethodHandleAccessor.java:104)
	at java.base/java.lang.reflect.Method.invoke(Method.java:565)
	at org.testng.internal.invokers.MethodInvocationHelper.invokeMethod(MethodInvocationHelper.java:141)
	at org.testng.internal.invokers.InvokeMethodRunnable.runOne(InvokeMethodRunnable.java:47)
	at org.testng.internal.invokers.InvokeMethodRunnable.call(InvokeMethodRunnable.java:76)
	at org.testng.internal.invokers.InvokeMethodRunnable.call(InvokeMethodRunnable.java:11)
	at java.base/java.util.concurrent.FutureTask.run(FutureTask.java:328)
	at java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1090)
	at java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:614)
	at java.base/java.lang.Thread.run(Thread.java:1474)

The test was measuring the delay from the moment it started a thread to shut down a bookie. However, lostBookieRecoveryDelay starts only after the auditor observes the lost bookie and schedules the delayed audit task.

Modifications

• Wait until the auditor has scheduled the delayed auditTask before starting the delay assertions.
• Keep the negative assertion anchored to the configured delay window, verifying that the ledger is not marked under-replicated before the delay expires.
• Use a short grace period after the delay window for the scheduled audit to run and for the under-replication watcher to observe the result.
• Add a helper to wait for the delayed audit task to be scheduled without triggering an audit directly.

Verifying this change

• Make sure that the change passes the CI checks.

Does this pull request potentially affect one of the following parts:

• Dependencies (add or upgrade a dependency)
• The public API
• The schema
• The default values of configurations
• The threading model
• The binary protocol
• The REST endpoints
• The admin CLI options
• The metrics
• Anything that affects deployment