If the node rejects a tx for any reason, chronik returns the error with bitcoind-rejected-tx in the error msg. Currently this is treated by chronik as a server issue, causing failoverProxy to cycle through all available nodes -- which will all throw the same error -- which will then return an unuseful 'chronik unavailable' msg to the user.
Do not move to the next chronik instance if you have a node rejection error and instead return that error.