[fix][test] Absorb Backoff jitter in PulsarServiceNameResolverTest.testRemoveUnavailableHost

[merlimat]

Motivation

The Backoff refactor in #25278 changed jitter from a one-sided "decrease only, clamped to initial" model to symmetric ±5%. As a result, the first call to Backoff.next() with initialDelay=1000ms can now return up to ~1050ms, whereas before it always returned exactly 1000ms.

PulsarServiceNameResolverTest.testRemoveUnavailableHost sleeps for exactly INIT_QUARANTINE_TIME_MS (1000ms) and then expects host1 to recover. With positive jitter, the recovery delay can exceed the sleep, so the elapsed-time check in computeEndpointStatus returns false and host1 stays unavailable, producing a flaky assertion failure on CI:

java.lang.AssertionError: Sets differ:
  expected [host2/<unresolved>:6651, host1/<unresolved>:6651, host3/<unresolved>:6651]
       but got [host2/<unresolved>:6651, host3/<unresolved>:6651]

Modifications

Sleep INIT_QUARANTINE_TIME_MS + 200ms instead of exactly INIT_QUARANTINE_TIME_MS, to absorb the ±5% jitter applied by the new Backoff implementation.

Verifying this change

This change is a trivial test fix.

Verified locally by running the test 5x with --rerun-tasks:

./gradlew :pulsar-client-original:test --tests \
  "org.apache.pulsar.client.impl.PulsarServiceNameResolverTest.testRemoveUnavailableHost" \
  --rerun-tasks

All 5 runs passed.

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

[lhotari]

LGTM