Add high-level testing of the pylogbook API by mocking out the low-level client with pre-canned responses

The strategy is basically: "assuming that the logbook responses continue to look as they do today, does the interface work".

This approach has already thrown up a few issues with the implementation that I wouldn't have found unless we did live testing (which isn't practical). Of course, our pre-canned responses will change over time, so we need to be mindful of that - it is for that reason that I've committed the script to build those responses.