Add a health endpoint to the code-mode gRPC listener (#38806)

## What changed

- Serve `GET /healthz` with a `200 OK` response over HTTP/1.1 and HTTP/2.
- Continue requiring HTTP/2 for all other requests so gRPC methods are not exposed over HTTP/1.1.

## Testing

- Add TCP listener integration coverage for HTTP/1.1 and HTTP/2 health checks and rejection of HTTP/1.1 gRPC requests.

GitOrigin-RevId: ae7bbf56323fbc76769375a6d8e90653e8adc860

codex-rs/code-mode-host/src/grpc_transport.rs

@@ -4,9 +4,17 @@ use std::net::SocketAddr;
 
 use anyhow::Context;
 use anyhow::Result;
+use axum::extract::Request;
+use axum::http::Method;
+use axum::http::StatusCode;
+use axum::http::Version;
+use axum::middleware;
+use axum::middleware::Next;
+use axum::routing::get;
 use codex_code_mode_protocol::grpc::code_mode_host_server::CodeModeHostServer;
 use codex_code_mode_protocol::host::MAX_FRAME_BYTES;
 use tokio::net::TcpListener;
+use tonic::service::Routes;
 use tonic::transport::Server;
 use tonic::transport::server::TcpIncoming;
 use tracing::info;
@@ -24,12 +32,28 @@ pub(super) async fn run_tcp_listener(bind_address: SocketAddr) -> Result<()> {
         .flush()
         .context("failed to publish code-mode gRPC listen address")?;
 
+    let routes = Routes::new(
+        CodeModeHostServer::new(GrpcCodeModeHost::new())
+            .max_decoding_message_size(MAX_FRAME_BYTES)
+            .max_encoding_message_size(MAX_FRAME_BYTES),
+    )
+    .into_axum_router()
+    .route("/healthz", get(|| async { StatusCode::OK }))
+    .layer(middleware::from_fn(
+        |request: Request, next: Next| async move {
+            if request.version() != Version::HTTP_2
+                && (request.method() != Method::GET || request.uri().path() != "/healthz")
+            {
+                return Err(StatusCode::HTTP_VERSION_NOT_SUPPORTED);
+            }
+
+            Ok(next.run(request).await)
+        },
+    ));
+
     Server::builder()
-        .add_service(
-            CodeModeHostServer::new(GrpcCodeModeHost::new())
-                .max_decoding_message_size(MAX_FRAME_BYTES)
-                .max_encoding_message_size(MAX_FRAME_BYTES),
-        )
+        .accept_http1(/*accept_http1*/ true)
+        .add_routes(routes.into())
         .serve_with_incoming(listener)
         .await
         .context("code-mode gRPC TCP listener failed")