In production, seeing instances where valid chronik queries may return unparseable errors if the server is re-indexing. In this case, we want to try a different server. But right now, chronik-client only recognizes the need to try a different server if the server is unreachable.
The unreachable server case is distinguished by 'code' in err. We add a new filter for "try another server" -- if we are unable to decode the error msg returned by the server according to chronik's proto.
If we can't parse the error, it's not a defined chronik error, and there is something wrong with the server.