fix(openai): gracefully handle stream endings

- The Responses API ends streams without a final '[DONE]' message.
- This causes reqwest_eventsource to return Error::StreamEnded.
- Previously, this was treated as a premature termination, triggering an error probe.
- We now explicitly match and break on Err(StreamEnded) for normal completion.

src/providers/openai.rs

@@ -524,6 +524,7 @@ impl super::Provider for OpenAIProvider {
                         }
                     }
                     Ok(_) => continue,
+                    Err(reqwest_eventsource::Error::StreamEnded) => break,
                     Err(e) => {
                         // Attempt to probe for the actual error body
                         let probe_resp = probe_client
@@ -887,6 +888,7 @@ impl super::Provider for OpenAIProvider {
                         }
                     }
                     Ok(_) => continue,
+                    Err(reqwest_eventsource::Error::StreamEnded) => break,
                     Err(e) => {
                          // Attempt to probe for the actual error body
                         let probe_resp = probe_client