Avoid occasional "runner has been closed" error during indexing

[jjezra]

As it seems:
Synchronized runner is created by runner.startSynchronizedSessionAsync, while the calling runner sets itself as the underlying runner (which is the main runner for the indexing session).
Let it be mainRunner and SyncRunner (where the latest points to `mainRunner).
runWithSynchronizedRunnerAndEndSession is called with the syncRunner, saves it in indexingCommon, the uses MoreAsyncUtil.composeWhenComplete to wait for the code to finish, then remove the syncRunner from indexingCommon.
When the indexer is closed, mainRunner is closed and if a syncRunner is not null, it closes too.
Typically the indexer is closed only after MoreAsyncUtil.composeWhenComplete gets to clear the syncRunner from common. But sometimes (seen it in exception path), the indexer is closed first and then:
Closes mainRunner
Closes syncRunner, which attempts to close the underlying runner, which is the mainRunner that cannot be re-closed.