Add documentation for clustered deployments

[ghost]

Some failures have been reported under clustered deployments, we need to clarify the documentation on the scenarios and how to use the scout agent appropriately within them:

• scout.listen() in the master and the middleware registration in the worker
• scout.listen() and middleware registration in the worker

The auto-creation of agent connections (when the middleware is called without an existing connection present) and how that interacts with clustered mode should also be explained if necessary.

[ghost]

It looks like this was solved in part by #261, scout had problems starting due to this -- there should actually be no extra documentation needed to solve cluster deployments as well, but it might be a good idea to at least show a normalized working setup.