fix(api): handle array message content on OpenAI chat completions (#15359)

### Related issues

Closes #15358

<!-- After filing upstream, replace XXXX with your issue number. -->

---

### What problem does this PR solve?

`POST /api/v1/openai/<chat_id>/chat/completions` forwards `messages` to
`async_chat` without normalizing `content`. Downstream, `dialog_service`
assumes string content:

```python
re.sub(r"##\d+\$\$", "", m["content"])
```

OpenAI-compatible clients may send `content` as an **array** of parts
(text, `image_url`, etc.), including text-only arrays. That causes
`TypeError` and HTTP **500** instead of a valid response or a clear
**400**.

`openai_api.py` also reads `messages[-1]["content"]` directly for
`prompt` without handling list-shaped content.

This PR normalizes array `content` to a string (concatenating `type:
text` parts) before calling `async_chat`, matching a minimal
OpenAI-compat path. Image parts can be documented as unsupported or
handled in a follow-up if vision integration is required.

test/testcases/test_http_api/test_session_management/test_session_sdk_routes_unit.py

@@ -1045,7 +1045,112 @@ def test_openai_nonstream_branch_unit(monkeypatch):
 
     res = _run(inspect.unwrap(module.openai_chat_completions)("chat-1"))
     assert res["choices"][0]["message"]["content"] == "world"
-    
+
+
+@pytest.mark.p2
+def test_openai_defaults_to_nonstream_when_stream_omitted_unit(monkeypatch):
+    """Omitted stream must default to false (OpenAI API compat), not SSE."""
+    module = _load_openai_api_module(monkeypatch)
+
+    monkeypatch.setattr(module, "num_tokens_from_string", lambda text: len(text or ""))
+    monkeypatch.setattr(
+        module.DialogService,
+        "query",
+        lambda **_kwargs: [SimpleNamespace(kb_ids=[], llm_id="chat-model", tenant_id="tenant-1")],
+    )
+
+    stream_flags = []
+
+    async def fake_async_chat(_dia, _msg, stream, **_kwargs):
+        stream_flags.append(stream)
+        yield {"answer": "hello", "reference": {}}
+
+    monkeypatch.setattr(module, "async_chat", fake_async_chat)
+    monkeypatch.setattr(
+        module,
+        "get_request_json",
+        lambda: _AwaitableValue(
+            {
+                "model": "model",
+                "messages": [{"role": "user", "content": "hi"}],
+            }
+        ),
+    )
+
+    res = _run(inspect.unwrap(module.openai_chat_completions)("chat-1"))
+    assert stream_flags == [False]
+    assert isinstance(res, dict)
+    assert res["object"] == "chat.completion"
+    assert res["choices"][0]["message"]["content"] == "hello"
+
+
+@pytest.mark.p2
+def test_openai_array_text_content_normalized_unit(monkeypatch):
+    """OpenAI-style array content with text parts must not crash async_chat."""
+    module = _load_openai_api_module(monkeypatch)
+
+    monkeypatch.setattr(module, "num_tokens_from_string", lambda text: len(text or ""))
+    monkeypatch.setattr(
+        module.DialogService,
+        "query",
+        lambda **_kwargs: [SimpleNamespace(kb_ids=[], llm_id="chat-model", tenant_id="tenant-1")],
+    )
+
+    captured_msg = []
+
+    async def fake_async_chat(_dia, msg, _stream, **_kwargs):
+        captured_msg.append(msg)
+        yield {"answer": "ok", "reference": {}}
+
+    monkeypatch.setattr(module, "async_chat", fake_async_chat)
+    monkeypatch.setattr(
+        module,
+        "get_request_json",
+        lambda: _AwaitableValue(
+            {
+                "model": "model",
+                "stream": False,
+                "messages": [
+                    {
+                        "role": "user",
+                        "content": [
+                            {"type": "text", "text": "Hello"},
+                            {"type": "text", "text": "World"},
+                        ],
+                    }
+                ],
+            }
+        ),
+    )
+
+    res = _run(inspect.unwrap(module.openai_chat_completions)("chat-1"))
+    assert captured_msg[0][0]["content"] == "Hello\nWorld"
+    assert res["choices"][0]["message"]["content"] == "ok"
+
+
+@pytest.mark.p2
+def test_openai_invalid_message_content_type_unit(monkeypatch):
+    module = _load_openai_api_module(monkeypatch)
+
+    monkeypatch.setattr(
+        module.DialogService,
+        "query",
+        lambda **_kwargs: [SimpleNamespace(kb_ids=[], llm_id="chat-model", tenant_id="tenant-1")],
+    )
+    monkeypatch.setattr(
+        module,
+        "get_request_json",
+        lambda: _AwaitableValue(
+            {
+                "model": "model",
+                "messages": [{"role": "user", "content": 12345}],
+            }
+        ),
+    )
+
+    res = _run(inspect.unwrap(module.openai_chat_completions)("chat-1"))
+    assert "messages[].content must be a string or an array of content parts." in res["message"]
+
 
 @pytest.mark.p2
 def test_agents_openai_compatibility_unit(monkeypatch):