Router unknown message handling callback registration is cleaner

Neater callback mechanism for device class instances to use. ...

This new architecture ensures default message processing always happens

Fixed bug where sensorid was always zero in csv log

Synthetic receive working in monitor_mihome.py

Updated TODO list with next steps for receive pipeline

energenie.loop() written with FSK receiver connected to fsk_router

Router ready to be tied into a loop()

control_both.py working in simulation

switch.py renamed as control_mihome.py, rewritten for new OO interface

legacy.py rewritten to use new device class interface

Tested on Raspberry Pi, all 4 testers still work with new Message()

Updated TODO list for the next phase of work on device_classes branch ...

Specifically re-testing samples, then pushing OO classes up into app.

Improved debug output for testing

Ran all test cases again (no hardware), fixed a few issues. ...

Need to still run this up on the hardware to test properly

Code now ready for regression testing with new Message() wrappers added

Refactored usages of raw message to use Message() ...

Removed all usages of showMessage, alterMessage

Marked with ##HERE markers the refactoring required for new Message() ...

Disabled alterMessage showMessage and createMessage to force this issue

OpenThings.Message tests now completed

PARAM_NAME key syntax for OpenThings.Message [set] tested ok

OpenThings.Message now tests attribute set and get. ...

Added decorators for @test_0 and @test_1 to turn tests on and off selectively

setattr tests done on OpenThings.Message

Tested all attribute read services in OpenThings.Message

OpenThings.Message test harness keyed paths tested and working

Characterised usage modes of OpenThings.Message better

Refinements to different parameter formats

Clarified test cases for OpenThings.Message()

Started writing unittest harness for new OpenThings.Message abstraction

Message.get() implemented and tested with paramid keying

Started to think through a better message abstraction

Started to think through a better alterMessage ...

characters that can be used in an identifier need checking