Update the frameless realtime default model (#40321)

## What changed

- Use `gpt-live-1-codex` as the default model for frameless realtime sessions.
- Preserve an explicitly supplied per-session model override.

## Testing

- Cover both default and explicit model selection in the WebRTC frameless conversation test.
- Update the app-server live session request expectation for the new default.

GitOrigin-RevId: ce5dc978911ff0289d677b60fae2dd8cb1058d22

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

@@ -1907,7 +1907,7 @@ async fn webrtc_v3_start_posts_live_session_and_joins_without_session_update() -
     assert_call_create_multipart(
         harness.call_capture.single_request(),
         "v=offer\r\n",
-        r#"{"audio":{"output":{"voice":"cove"}},"delegation":{"ack_filler":false,"type":"client"},"instructions":"backend prompt\n\nstartup context","model":"gpt-live-1-boulder-alpha"}"#,
+        r#"{"audio":{"output":{"voice":"cove"}},"delegation":{"ack_filler":false,"type":"client"},"instructions":"backend prompt\n\nstartup context","model":"gpt-live-1-codex"}"#,
         "/v1/live",
     )?;
     assert!(