Wait for requested delay in rate limit errors (#2266)

Fixes: https://github.com/openai/codex/issues/2131

Response doesn't have the delay in a separate field (yet) so parse the
message.

codex-rs/core/src/codex.rs

@@ -1281,7 +1281,10 @@ async fn run_turn(
                 let max_retries = sess.client.get_provider().stream_max_retries();
                 if retries < max_retries {
                     retries += 1;
-                    let delay = backoff(retries);
+                    let delay = match e {
+                        CodexErr::Stream(_, Some(delay)) => delay,
+                        _ => backoff(retries),
+                    };
                     warn!(
                         "stream disconnected - retrying turn ({retries}/{max_retries} in {delay:?})...",
                     );
@@ -1391,6 +1394,7 @@ async fn try_run_turn(
             // Treat as a disconnected stream so the caller can retry.
             return Err(CodexErr::Stream(
                 "stream closed before response.completed".into(),
+                None,
             ));
         };
 
@@ -2201,6 +2205,7 @@ async fn drain_to_completed(sess: &Session, sub_id: &str, prompt: &Prompt) -> Co
         let Some(event) = maybe_event else {
             return Err(CodexErr::Stream(
                 "stream closed before response.completed".into(),
+                None,
             ));
         };
         match event {
@@ -2218,6 +2223,7 @@ async fn drain_to_completed(sess: &Session, sub_id: &str, prompt: &Prompt) -> Co
                     None => {
                         return Err(CodexErr::Stream(
                             "token_usage was None in ResponseEvent::Completed".into(),
+                            None,
                         ));
                     }
                 };