* physical/spanner: use separate client for updating locks
We believe this mitigates an issue where a large influx of requests
cause the leader to be unable to update the lock table (since it cannot
grab a client from the pool or the client has no more open connections),
which causes cascading failure.
* storage/gcs: fix race condition in releasing lock
Previously we were deleting a lock without first checking if the lock we were deleting was our own. There existed a small period of time where vault-0 would lose leadership and vault-1 would get leadership. vault-0 would delete the lock key while vault-1 would write it. If vault-0 won, there'd be another leader election, etc.
This fixes the race by using a CAS operation instead.
* storage/gcs: properly break out of loop during stop
* storage/spanner: properly break out of loop during stop
Josh Black