Implementation of log handler

[wcpettus]

In dl2, the python code provided "log handlers", which were initialized with every service:
https://github.com/project8/dragonfly/blob/dl2/master/bin/dragonfly#L103
And parsed the log messages and could provide action for any log message that exceeded a threshold value. In practice, this meant broadcasting logs that exceeded a threshold to the alerts exchange:
https://github.com/project8/dragonfly/blob/dl2/master/dragonfly/status_log_handlers/amqp_handler.py#L53
(where the slack service could push notifications to the appropriate slack channels)

In dl3, with logging integrated between cpp and python, this properly would need to move back into the cpp code to replicate.

[nsoblath]

Initial thoughts: since I reimplemented the logger setup using the Quill logging library, we could implement functionality similar to the log handler with a custom sink.

This would be really nice from the P8 Psyllid/Sandfly perspective because I could remove the message relayer if sending of status_messages is implemented within the logger.

More technical thoughts:

• In the write_log() function, the log level is one of the arguments. This would be used in a way similar to what's done in the dl2 amqp_handler.
• The constructor for the sink can take arguments that are passed to the constructor by optional arguments in create_or_get_sink<sink_type>().
• The actual message sending can be done as in Sandfly's message_relayer.

[nsoblath]

Updating this for the spdlog-based logger setup.

This should be done by implementing an spdlog sink, as described here.

On the dl side, we use the relayer class to send the messages.