fix(iroh): Try to fix flaky test_token_passthrough test (#1419)

## Description

This seems to fail often on CI on the emulated platforms when the
machines are busy.  And the failures we see are timeouts.  So very
likely increasing this timeout should help.

Does a few smaller fixes:

- Only bind on localhost instead of exposing us to the world.  Tests
  should not expose themselves to the world.

- Make sure that the receiver dropping is caught in a panic.

## Notes & open questions

## Change checklist

- [x] Self-review.
- [x] ~~Documentation updates if relevant.~~
- [x] Tests if relevant.

iroh/tests/provide.rs

@@ -877,7 +877,7 @@ async fn test_token_passthrough() -> Result<()> {
     let rt = test_runtime();
     let expected = b"hello".to_vec();
     let (db, hash) = create_test_db([("test", expected.clone())]);
-    let addr = "0.0.0.0:0".parse().unwrap();
+    let addr = SocketAddr::from((Ipv4Addr::LOCALHOST, 0));
     let node = test_node(db, addr)
         .custom_auth_handler(Arc::new(CustomAuthHandler))
         .runtime(&rt)
@@ -894,7 +894,7 @@ async fn test_token_passthrough() -> Result<()> {
                     if let iroh_bytes::provider::Event::GetRequestReceived { token: tok, .. } =
                         bp_msg
                     {
-                        events_sender.send(tok).ok();
+                        events_sender.send(tok).expect("receiver dropped");
                     }
                 }
             }
@@ -905,7 +905,7 @@ async fn test_token_passthrough() -> Result<()> {
 
     let addrs = node.local_endpoint_addresses().await?;
     let peer_id = node.peer_id();
-    tokio::time::timeout(Duration::from_secs(10), async move {
+    tokio::time::timeout(Duration::from_secs(30), async move {
         let endpoint = MagicEndpoint::builder()
             .secret_key(SecretKey::generate())
             .keylog(true)