Hive tasks receive statelog updates from unauthorized slaves

[jkarder]

It is possible that two slaves are calculating the same task at the same time, e.g. when the first slave looses network connectivity, the task will be rescheduled on another slave. If after a while the first slave can reconnect to the Hive service, it is possible that this slave also updates statelog entries. This can cause the server to abort both calculations and result in an aborted task which is not rescheduled.

[jkarder]

r17575: fixed statelog updates

• fixed UpdateTasks so that it selects the correct last slave id for a specified task
• introduced check if a slave is allowed to update a task's state (IsAuthorizedForTask)