fix(responses): keep raw reasoning on content only

Raw chain-of-thought was written to both `content` (`reasoning_text`) and
`summary` (`summary_text`), and the stream emitter sent the same delta on
`response.reasoning_text.delta` *and* `response.reasoning_summary_text.delta`.

Clients that render both channels therefore printed every thinking chunk
twice — most visibly the Codex CLI, whose thinking panel repeated itself.

OpenAI keeps the two channels distinct: `content` carries the raw CoT while
`summary` is the summarised view. Emit the thinking on `content` only:

- `openai_responses_reasoning_text_fields` becomes
  `openai_responses_reasoning_text_parts`, returning just the `content` array;
  reasoning items keep `summary: []` (or a provider-supplied summary).
- The Responses stream emitter emits `response.reasoning_text.delta` /
  `.done` and no longer mirrors them onto the summary events. The reasoning
  `output_item.added` no longer announces a `reasoning_summary_part`.
- The provider-state reasoning reader accepts `content` (`reasoning_text`)
  first and falls back to `summary`, so it also understands items produced by
  older Aether versions; its state field is renamed accordingly.
- Non-streaming builders (Chat -> Responses, manual Responses response, Grok
  gateway) place the thinking on `content` and leave `summary` empty.

Tests cover the raw thinking appearing exactly once in the emitted stream.

apps/aether-gateway/src/execution_runtime/grok.rs

@@ -3203,10 +3203,7 @@ fn openai_responses_body(
             "id": openai_responses_synthetic_reasoning_item_id(&response_id, 0),
             "type": "reasoning",
             "status": "completed",
-            "summary": [{
-                "type": "summary_text",
-                "text": thinking,
-            }],
+            "summary": [],
             "content": [{
                 "type": "reasoning_text",
                 "text": thinking,
@@ -4640,10 +4637,7 @@ mod tests {
             body["output"][0]["content"][0]["text"],
             serde_json::json!("short reasoning")
         );
-        assert_eq!(
-            body["output"][0]["summary"][0]["text"],
-            serde_json::json!("short reasoning")
-        );
+        assert_eq!(body["output"][0]["summary"], serde_json::json!([]));
         assert_eq!(body["output"][1]["type"], serde_json::json!("message"));
         assert!(body["output"][1]["id"]
             .as_str()
@@ -4827,7 +4821,12 @@ mod tests {
 
         assert!(body.contains("event: response.created"));
         assert!(body.contains("event: response.in_progress"));
-        assert!(body.contains("event: response.reasoning_summary_part.added"));
+        // Thinking must stay off the summary channel or clients that render
+        // both (Codex) print the raw chain-of-thought twice.
+        assert!(!body.contains("event: response.reasoning_summary_part.added"));
+        assert!(!body.contains("event: response.reasoning_summary_text.delta"));
+        assert!(!body.contains("event: response.reasoning_summary_text.done"));
+        assert!(body.contains("\"type\":\"reasoning_text\""));
         assert!(body.contains("event: response.content_part.added"));
         assert!(body.contains("event: response.output_text.done"));
         assert!(body.contains("event: response.completed"));

apps/aether-gateway/src/execution_runtime/stream/execution.rs

@@ -11880,7 +11880,11 @@ mod tests {
             .expect("response body should read");
         let body = String::from_utf8(body.to_vec()).expect("response body should be utf8");
         assert!(
-            body.contains("event: response.reasoning_summary_text.delta\n"),
+            body.contains("event: response.reasoning_text.delta\n"),
+            "{body}"
+        );
+        assert!(
+            !body.contains("event: response.reasoning_summary_text.delta\n"),
             "{body}"
         );
         assert!(