Shutdown Synchronization #1412

The remainder of the panics in the test suite are most likely due to the shutdown of the node and its services not being synchronized.

Currently the shutdown of the node is not blocking. It cancels the context and calls some close methods but it doesn't block and wait to confirm that the services were in fact shutdown.

Blocking on shutdown would have surfaced many of the issues that were recently fixed with the context handling.

This can be done with either a shutdown chan or a waitgroup that is signaled when launching go routines and closing services.

This will ensure that the node only returns from it's stop call once all services have stopped, thus preventing any panics due to writing to a closed file.