fix(provider): retry Codex server_error

Mid-stream response.failed events from OpenAI carry
'server_error' (underscore), but the transient-error marker
list only had 'server error' (space). is_transient_response
missed it, so the retry loop gave up on the first attempt
instead of using its normal 1s/2s/4s backoff.

nanobot/providers/base.py

@@ -326,6 +326,7 @@ class LLMProvider(ABC):
         "timed out",
         "connection",
         "server error",
+        "server_error",
         "temporarily unavailable",
         "速率限制",
         "访问量过大",

tests/providers/test_openai_codex_provider.py

@@ -319,6 +319,25 @@ async def test_codex_timeout_error_is_typed_and_retryable(monkeypatch) -> None:
     assert response.error_should_retry is True
 
 
+@pytest.mark.asyncio
+async def test_codex_mid_stream_server_error_is_treated_as_transient(monkeypatch) -> None:
+    _mock_codex_token(monkeypatch)
+
+    async def fake_request(*args, **kwargs):
+        raise RuntimeError(
+            "Response failed: {'type': 'server_error', 'code': 'server_error', "
+            "'message': 'An error occurred while processing your request.'}"
+        )
+
+    monkeypatch.setattr("nanobot.providers.openai_codex_provider._request_codex", fake_request)
+
+    provider = OpenAICodexProvider()
+    response = await provider.chat([{"role": "user", "content": "hello"}])
+
+    assert response.finish_reason == "error"
+    assert provider_base.LLMProvider.is_transient_response(response) is True
+
+
 @pytest.mark.asyncio
 async def test_codex_provider_passes_proxy_to_oauth_and_response_request(monkeypatch) -> None:
     proxy = "http://127.0.0.1:23458"