Handling of cancellation of ModelController operations is broken

When deploying via the the FS deployment scanner a timeout causes all subsequent deployments to fail with further timeout failures.

The problem appears to be that FileSystemDeploymentService submits the DeployTask for timed execution it acquires a lock it's copy of ContainerStateMonitor (via OperationContextImpl.acquireContainerMonitor()). When a timeout exception is throw by the first failure this lock is never cleaned up which results in all further deployment jobs to hang on ContainerStateMonitor.await (line 161).

Restarting the server clears the monitor and allows the scanner to be reused as normal.