Allow for tasks/flows to be automatically timed out

A task/flow now once started can run forever, it would be useful to provide a simple mechanism for interrupting tasks which have exceeded there allotted time (or if not possible, allow the flow to be timed out itself, this would automatically stop future task execution).

Heat may be a useful example of this (although the implementation will vary for taskflow):

- https://github.com/openstack/heat/blob/2013.2.rc1/heat/engine/scheduler.py#L49

Since taskflow is not (natively) co-routine based (if an eventlet executor is used, we do automatically become co-routine based) we would have to create a slightly different mechanism to trigger a task timeout (heat can just force the task to get an exception when the task yields next).