Introduce and refine SocketDevice

Experiments with a device that does real communication instead of simulation look promising. I think it's worthwhile spending some time trying to improve the design (or actually design something in the first place) so that it becomes easy to create such devices.

Owen pointed out that the name `SocketDevice` might not be appropriate, but I can't come up with a better name, so suggestions are welcome.

This is rather low on the list of priorities, so I'm putting the backlog label on it.