Normally user should not been able to complete the task (of sub-process) after t

To the best of my knowledge, the behavior you were observing could be because of the following reasons:
1. A lag time between the completion of related action and starting the termination of the process instance (probably via messaging or some other means).
2. A lag time between the activities that should be run before termination of the instance and termination of instance.

In general, as and when the process is terminated, within a fraction of seconds you can observe the synchronous sub-processes (and obviously their tasks) being terminated.

If you start terminating the process from back end immediately after the completion of related action (immediately after submission of the task) and further if the default set of activities that are run before termination of actual instance are completed in no time , I hope definitely you won't be observing any more scenarios as such specified by you.

To the best of my knowledge, the behavior you were observing could be because of the following reasons:
1. A lag time between the completion of related action and starting the termination of the process instance (probably via messaging or some other means).
2. A lag time between the activities that should be run before termination of the instance and termination of instance.

In general, as and when the process is terminated, within a fraction of seconds you can observe the synchronous sub-processes (and obviously their tasks) being terminated.

If you start terminating the process from back end immediately after the completion of related action (immediately after submission of the task) and further if the default set of activities that are run before termination of actual instance are completed in no time , I hope definitely you won't be observing any more scenarios as such specified by you.