Performance Refactoring

[almgong]

• We really only need one separate Thread to run both the discovery and probe tasks. However, something more promising is using ScheduledThreahPoolExecutor, which exposes functions to run Runnable objects every 'x' period of time.
• Application-wise, NDN communication averages about 50-150ms. The delay between processing events is set to 50. Is this normal?

[almgong]

First point resolved using ScheduledThreadPoolExecutor in 5af9a33.

[almgong]

Ideally, all tasks should eventually get converted to runnables, and the thread pool executor should be invoked. This way we can actually limit the number of threads created during operation.However tasks such as registering prefixes should NOT be a runnable, as they may never terminate or will do so after a long time. Tasks that are short lived, such as the FaceCreateTask are prime targets for this refactoring.