Add opt-in concurrent exec-server request dispatch (#36987)

## Why

Sequential dispatch lets a long-running request block unrelated health checks
and cleanup on the same connection.

## What changed

- Add `--concurrent-requests <COUNT>` for local and remote exec-server
  connections, while retaining sequential dispatch when the option is omitted
  or set to `1`.
- Preserve handshake ordering before enabling concurrent dispatch.
- Reserve separate capacity for status, signal, terminate, and close requests so
  they remain responsive when ordinary request capacity is saturated.
- Drain queued client responses during disconnect and cancel outstanding
  request tasks during connection shutdown.

## Testing

- Cover CLI parsing and concurrency-limit validation.
- Verify default sequential behavior, pipelined handshake ordering, concurrent
  request progress, control-request responsiveness, and disconnect handling.

GitOrigin-RevId: 48e4b092e318204ed635543f01f9ee0e7df095fc

codex-rs/cli/src/main.rs

@@ -573,6 +573,14 @@ struct ExecServerCommand {
     #[arg(long = "strict-config", default_value_t = false)]
     strict_config: bool,
 
+    /// Maximum number of requests to process concurrently on each connection.
+    #[arg(
+        long = "concurrent-requests",
+        value_name = "COUNT",
+        default_value = "1"
+    )]
+    request_dispatch_mode: codex_exec_server::RequestDispatchMode,
+
     /// Transport endpoint URL. Supported values: `ws://IP:PORT` (default), `stdio`, `stdio://`.
     #[arg(long = "listen", value_name = "URL", conflicts_with = "remote")]
     listen: Option<String>,
@@ -1745,6 +1753,7 @@ async fn run_exec_server_command(
         if let Some(name) = cmd.name {
             remote_config.name = name;
         }
+        remote_config.request_dispatch_mode = cmd.request_dispatch_mode;
         let remote_config = remote_config.with_telemetry(telemetry);
         let parent_lifetime = if cmd.exit_on_stdin_close {
             exec_server_telemetry::ParentLifetime::StdinPipe
@@ -1794,6 +1803,7 @@ async fn run_exec_server_command(
                     runtime_paths,
                     telemetry,
                     http_client_factory,
+                    cmd.request_dispatch_mode,
                 )
                 .await
             },

codex-rs/cli/tests/exec_server.rs

@@ -90,6 +90,24 @@ fn local_exec_server_ignores_invalid_config_without_strict_config() -> Result<()
     Ok(())
 }
 
+/// The standalone exec-server accepts an explicit per-connection concurrency limit.
+#[test]
+fn local_exec_server_accepts_concurrent_requests_flag() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    let mut cmd = codex_command(codex_home.path())?;
+    cmd.args([
+        "exec-server",
+        "--listen",
+        "stdio",
+        "--concurrent-requests",
+        "2",
+    ])
+    .assert()
+    .success();
+
+    Ok(())
+}
+
 #[test]
 fn local_exec_server_allows_disabled_parent_lifetime_environment_variable() -> Result<()> {
     let codex_home = TempDir::new()?;
@@ -173,6 +191,8 @@ metrics_exporter = {{ otlp-http = {{ endpoint = "{collector_url}/v1/metrics", pr
             &registry.uri(),
             "--environment-id",
             ENVIRONMENT_ID,
+            "--concurrent-requests",
+            "2",
         ])
         .stdin(Stdio::piped())
         .stderr(Stdio::piped())