On one side, the ShardingDDLCoordinator is creating and destroying several OperationContext objects since it uses a task chain to perform its operations. On the other side, the lock state is bound to an OperationContext through a Locker object, and that Locker object is created and destructed together with the OperationContext.

However, the new DDLLockManager implementation requires keeping a Locker object alive and attached to a ShardingDDLCoordinator in order to hold DDL locks across a task chain.

The aim of this ticket is to keep a unique Locker object per ShardingDDLCoordinator by attaching and detaching it to the current OperationContext once its constructed and destructed respectively.