Defer persistence of rollout file (#11028)

- Defer rollout persistence for fresh threads (`InitialHistory::New`):
keep rollout events in memory and only materialize rollout file + state
DB row on first `EventMsg::UserMessage`.
- Keep precomputed rollout path available before materialization.
- Change `thread/start` to build thread response from live config
snapshot and optional precomputed path.
- Improve pre-materialization behavior in app-server/TUI: clearer
invalid-request errors for file-backed ops and a friendlier `/fork` “not
ready yet” UX.
- Update tests to match deferred semantics across
start/read/archive/unarchive/fork/resume/review flows.
- Improved resilience of user_shell test, which should be unrelated to
this change but must be affected by timing changes

For Reviewers:
* The primary change is in recorder.rs
* Most of the other changes were to fix up broken assumptions in
existing tests

Testing:
* Manually tested CLI
* Exercised app server paths by manually running IDE Extension with
rebuilt CLI binary
* Only user-visible change is that `/fork` in TUI generates visible
error if used prior to first turn

codex-rs/app-server/tests/suite/v2/thread_resume.rs

@@ -35,7 +35,7 @@ const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs
 const CODEX_5_2_INSTRUCTIONS_TEMPLATE_DEFAULT: &str = "You are Codex, a coding agent based on GPT-5. You and the user share the same workspace and collaborate to achieve the user's goals.";
 
 #[tokio::test]
-async fn thread_resume_returns_original_thread() -> Result<()> {
+async fn thread_resume_rejects_unmaterialized_thread() -> Result<()> {
     let server = create_mock_responses_server_repeating_assistant("Done").await;
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri())?;
@@ -57,24 +57,26 @@ async fn thread_resume_returns_original_thread() -> Result<()> {
     .await??;
     let ThreadStartResponse { thread, .. } = to_response::<ThreadStartResponse>(start_resp)?;
 
-    // Resume it via v2 API.
+    // Resume should fail before the first user message materializes rollout storage.
     let resume_id = mcp
         .send_thread_resume_request(ThreadResumeParams {
             thread_id: thread.id.clone(),
             ..Default::default()
         })
         .await?;
-    let resume_resp: JSONRPCResponse = timeout(
+    let resume_err: JSONRPCError = timeout(
         DEFAULT_READ_TIMEOUT,
-        mcp.read_stream_until_response_message(RequestId::Integer(resume_id)),
+        mcp.read_stream_until_error_message(RequestId::Integer(resume_id)),
     )
     .await??;
-    let ThreadResumeResponse {
-        thread: resumed, ..
-    } = to_response::<ThreadResumeResponse>(resume_resp)?;
-    let mut expected = thread;
-    expected.updated_at = resumed.updated_at;
-    assert_eq!(resumed, expected);
+    assert!(
+        resume_err
+            .error
+            .message
+            .contains("no rollout found for thread id"),
+        "unexpected resume error: {}",
+        resume_err.error.message
+    );
 
     Ok(())
 }
@@ -322,6 +324,27 @@ async fn thread_resume_prefers_path_over_thread_id() -> Result<()> {
     .await??;
     let ThreadStartResponse { thread, .. } = to_response::<ThreadStartResponse>(start_resp)?;
 
+    let turn_id = mcp
+        .send_turn_start_request(TurnStartParams {
+            thread_id: thread.id.clone(),
+            input: vec![UserInput::Text {
+                text: "materialize".to_string(),
+                text_elements: Vec::new(),
+            }],
+            ..Default::default()
+        })
+        .await?;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(turn_id)),
+    )
+    .await??;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("turn/completed"),
+    )
+    .await??;
+
     let thread_path = thread.path.clone().expect("thread path");
     let resume_id = mcp
         .send_thread_resume_request(ThreadResumeParams {
@@ -339,9 +362,8 @@ async fn thread_resume_prefers_path_over_thread_id() -> Result<()> {
     let ThreadResumeResponse {
         thread: resumed, ..
     } = to_response::<ThreadResumeResponse>(resume_resp)?;
-    let mut expected = thread;
-    expected.updated_at = resumed.updated_at;
-    assert_eq!(resumed, expected);
+    assert_eq!(resumed.id, thread.id);
+    assert_eq!(resumed.path, thread.path);
 
     Ok(())
 }
@@ -412,12 +434,17 @@ async fn thread_resume_accepts_personality_override() -> Result<()> {
     skip_if_no_network!(Ok(()));
 
     let server = responses::start_mock_server().await;
-    let body = responses::sse(vec![
+    let first_body = responses::sse(vec![
         responses::ev_response_created("resp-1"),
         responses::ev_assistant_message("msg-1", "Done"),
         responses::ev_completed("resp-1"),
     ]);
-    let response_mock = responses::mount_sse_once(&server, body).await;
+    let second_body = responses::sse(vec![
+        responses::ev_response_created("resp-2"),
+        responses::ev_assistant_message("msg-2", "Done"),
+        responses::ev_completed("resp-2"),
+    ]);
+    let response_mock = responses::mount_sse_sequence(&server, vec![first_body, second_body]).await;
 
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri())?;
@@ -438,9 +465,30 @@ async fn thread_resume_accepts_personality_override() -> Result<()> {
     .await??;
     let ThreadStartResponse { thread, .. } = to_response::<ThreadStartResponse>(start_resp)?;
 
+    let materialize_id = mcp
+        .send_turn_start_request(TurnStartParams {
+            thread_id: thread.id.clone(),
+            input: vec![UserInput::Text {
+                text: "seed history".to_string(),
+                text_elements: Vec::new(),
+            }],
+            ..Default::default()
+        })
+        .await?;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(materialize_id)),
+    )
+    .await??;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("turn/completed"),
+    )
+    .await??;
+
     let resume_id = mcp
         .send_thread_resume_request(ThreadResumeParams {
-            thread_id: thread.id.clone(),
+            thread_id: thread.id,
             model: Some("gpt-5.2-codex".to_string()),
             personality: Some(Personality::Friendly),
             ..Default::default()
@@ -451,11 +499,11 @@ async fn thread_resume_accepts_personality_override() -> Result<()> {
         mcp.read_stream_until_response_message(RequestId::Integer(resume_id)),
     )
     .await??;
-    let _resume: ThreadResumeResponse = to_response::<ThreadResumeResponse>(resume_resp)?;
+    let resume: ThreadResumeResponse = to_response::<ThreadResumeResponse>(resume_resp)?;
 
     let turn_id = mcp
         .send_turn_start_request(TurnStartParams {
-            thread_id: thread.id,
+            thread_id: resume.thread.id,
             input: vec![UserInput::Text {
                 text: "Hello".to_string(),
                 text_elements: Vec::new(),
@@ -475,7 +523,10 @@ async fn thread_resume_accepts_personality_override() -> Result<()> {
     )
     .await??;
 
-    let request = response_mock.single_request();
+    let requests = response_mock.requests();
+    let request = requests
+        .last()
+        .expect("expected request for resumed thread turn");
     let developer_texts = request.message_input_texts("developer");
     assert!(
         developer_texts