fix(core): self-heal etcd meta watch on transport reconnect

[dpol1]

Purpose of the PR

• close [Improvement] Re-register HStore schema cache clear listener after a Meta reconnect #3036

CachedSchemaTransactionV2 registers a JVM-global watch on the meta store so that a schema change on one node clears the schema cache on the other nodes. That watch goes through EtcdMetaDriver.listen / listenPrefix, which handed jetcd the bare Consumer<WatchResponse> overload. That overload discards onError and onCompleted, so when jetcd ends a watch (e.g. after a transport reconnect tears down the gRPC stream) the listener stopped receiving events with no log line and no exception. The node kept serving stale schema and nothing reported it. PdMetaDriver does not have this problem because its KvClient already re-subscribes on error.

Main Changes

• EtcdMetaDriver.listen / listenPrefix now register a Watch.Listener instead of the bare Consumer overload, surfacing both onError and onCompleted.
  • onError only logs at WARN level. jetcd 0.5.9 (WatchImpl.WatcherImpl.handleError) already retries recoverable errors internally by notifying onError and rescheduling resume() on the same watcher; re-subscribing from onError would therefore open a duplicate watch on every transient reconnect.
  • onCompleted schedules a re-subscribe after a 1 s backoff on a daemon thread. onCompleted is jetcd's terminal-close signal: the old watcher is already removed and will not recover, so a replacement watch is safe and necessary. This mirrors the self-heal PdMetaDriver already gets from KvClient.
  • If the re-subscribe itself throws (e.g. the endpoint is still unreachable), it is retried with the same backoff instead of abandoning recovery after a single failure.
• Removed CachedSchemaTransactionV2.resetMetaListenerForReconnect(). It was a manual stopgap added in fix(server): sync hstore schema cache clears #3011 with no callers, and it never shipped in a tagged release. Now that the driver self-heals, the JVM-global register-once flag is correct to stay set; the manual reset has nothing left to do. The lifecycle comment on metaEventListenerRegistered is updated to describe the self-heal behaviour.
• The MetaDriver interface is unchanged, so neither implementor needs edits beyond EtcdMetaDriver.

Known limitation: the re-subscribe opens a fresh watch without a stored revision, so any cache-clear events emitted during the short reconnect window are not replayed. This is the same behaviour as PdMetaDriver / KvClient. The change removes the permanent silent failure; it does not add gap replay.

Verifying these changes

• Need tests and can be verified as follows:
  • New EtcdMetaDriverTest (JUnit + Mockito) mocks the jetcd Client/Watch via the package-private test constructor (no live etcd needed) and asserts:
    1. onCompleted triggers a re-subscribe (a second watch(...) call).
    2. onError does not trigger a re-subscribe (only logs), so no duplicate watch is opened.
    3. The re-created prefix watch preserves the original key and isPrefix WatchOption.
    4. onNext still delivers events to the consumer.
    5. A re-watch that throws once is retried and eventually succeeds (no permanent loss).
  • Local run: EtcdMetaDriverTest 5/5, CachedSchemaTransactionTest 18/18, MetaManagerSchemaCacheClearEventTest 6/6 after the stopgap removal.
  • Red-green: commenting out the re-subscribe in onCompleted makes the recovery tests fail; restoring it makes them pass.

Does this PR potentially affect the following parts?

• Dependencies
• Modify configurations
• The public API
• Other affects (typed here)
• Nope

Documentation Status

• Doc - TODO
• Doc - Done
• Doc - No Need

[imbajin]

Blocking issues found in the etcd watch recovery path: retryable jetcd watch errors can now create duplicate active watchers, and a failed delayed re-watch attempt can stop recovery permanently. Please fix these before merging; I also left one non-blocking test-hardening note.

[imbajin]

Blocking: no. Summary: No obvious issues found in the current head. Evidence: git diff --check origin/master...HEAD; EtcdMetaDriverTest passed 5 tests.

Non-blocking follow-ups:

• Please update the PR description to match the latest implementation: onError now only logs because jetcd retries recoverable errors itself, while re-watch is scheduled from terminal onCompleted; the test count is now 5, not 4.
• Consider reducing recoverable watch onError logging from WARN with stack trace to INFO/DEBUG or a rate-limited log, since jetcd is expected to self-retry those transient failures.
• If HugeGraph treats public core methods as a compatibility surface, consider keeping CachedSchemaTransactionV2.resetMetaListenerForReconnect() as a deprecated no-op or documented fallback. Given it was an uncalled stopgap that did not ship in a tagged release, I do not consider this blocking.