error handling

As mentioned in #44 , we would raise a python exception if things break. I've implemented support for working in a read-only state (configurable), and have caught these exceptions to log error messages that should be more helpful to the user. I've also taken the opportunity to handle verifying the configuration file, so attempting to reload a running service after breaking the config should keep running. I haven't implemented any specific probe-config command, do you think we should? Closes #44 PS: I haven't looked into LACT yet, no updates on that note.