The restart and cancel tasks functionalities are inherently limited and unsafe. They do not work on short-lived tasks and require the tasks to have a unique ID.
Furthermore, if 2 tasks have the same ID, the reulst is unpredicatble.
This means we should remove the features from the admin console, from the node admin mbean, from the NodeState object.
We should also deprecate the JPPFTask.onRestart() callback. We will keep JPPFTask.onCancel() because it is called on all the tasks when a job is cancelled, but its semantic will change and it must be made clear in the API specs and in the documentation.