Function spawn_tcp_bridge 

pub fn spawn_tcp_bridge(
    source_stage: usize,
    input_rx: Receiver<TransportBatch>,
    output_tx: SyncSender<TransportBatch>,
    config: TcpTransportConfig,
    listener: BridgeListener,
    addr: BridgeAddr,
    circuit_breaker: Option<CircuitBreaker>,
) -> JoinHandle<BridgeAccumulator>

Spawn a TCP transport bridge between pipeline stages.

In a multi-node deployment, the bind_addr and connect_addr can point to different physical interfaces or remote hosts.

circuit_breaker is optional and persists failure history across calls (unlike config.reconnect_attempts, which only bounds retries within this one call) — pass ClusterState::circuit_breaker_for(node_id) for a real remote node so a chronically-unreachable node fails fast on subsequent pipeline runs instead of repeating the full connect/backoff sequence every time. None (e.g. loopback benchmarking, where there’s no real remote failure mode to protect against) reproduces the prior always-retry behavior exactly.