Look into implementing/using a threaded dns resolver to make nc_resolve non-blocking

[TysonAndre]

In addition to a connecting state, maybe add a resolving state prior to that for server connections.

getaddrinfo is blocking - some applications (e.g. curl?) will start a separate thread and call getaddrinfo in that thread to avoid blocking the main thread.

Related to #40 - if there's local dns result caching on the host itself this is less of an issue but may still take milliseconds and be called repeatedly.

The heartbeat patches may be interacting with the server retry timeout poorly?