Restructuring channel power state to an enum

[ryan-summers]

This PR fixes #174 by refactoring the power state of channels to be a 3-valued enum.

TODO:

• Verify that the new format can be loaded from old EEPROM data.

[jordens]

Looks good.
There is no static analysis of the event/state combinations in smlang, right? I.e. if code does handle_event(X) while in state Y, we get a panic if there is no corresponding transition. Or does it ignore these events?

[ryan-summers]

When you call process_event(event), it returns a Result indicating whether or not the event was accepted for the given state. In some cases, we don't care if the transition failed, so we just ok() it. In others when we can guarantee that the transition will always be valid, we do an unwrap() instead since it should never panic.