fix: Send Completed event when stream ends without finish_reason

- Add debug logging for finish_reason to diagnose stream completion issues
- When SSE stream ends without sending a finish_reason chunk, emit final items and Completed event
- This fixes the "working" hang when the API stream closes without proper completion signal
- Fixed duplicate max_tokens fields in test provider definitions

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

llmx-rs/core/src/chat_completions.rs

@@ -841,6 +841,7 @@ async fn process_chat_sse<S>(
 
             // Emit end-of-turn when finish_reason signals completion.
             if let Some(finish_reason) = choice.get("finish_reason").and_then(|v| v.as_str()) {
+                debug!("Received finish_reason: {}", finish_reason);
                 match finish_reason {
                     "tool_calls" if fn_call_state.active => {
                         // First, flush the terminal raw reasoning so UIs can finalize
@@ -888,6 +889,22 @@ async fn process_chat_sse<S>(
             }
         }
     }
+
+    // Stream ended without finish_reason - this can happen when the stream closes abruptly
+    debug!("Stream ended without finish_reason, emitting final items and Completed event");
+    if let Some(item) = assistant_item.take() {
+        let _ = tx_event.send(Ok(ResponseEvent::OutputItemDone(item))).await;
+    }
+    if let Some(item) = reasoning_item.take() {
+        let _ = tx_event.send(Ok(ResponseEvent::OutputItemDone(item))).await;
+    }
+    // Send Completed event so llmx knows the turn is done
+    let _ = tx_event
+        .send(Ok(ResponseEvent::Completed {
+            response_id: String::new(),
+            token_usage: token_usage.clone(),
+        }))
+        .await;
 }
 
 /// Optional client-side aggregation helper

llmx-rs/core/src/model_provider_info.rs

@@ -295,7 +295,6 @@ pub fn built_in_model_providers() -> HashMap<String, ModelProviderInfo> {
                 stream_max_retries: None,
                 stream_idle_timeout_ms: None,
             max_tokens: None,
-                max_tokens: None,
                 requires_openai_auth: false,
             },
         ),
@@ -337,7 +336,6 @@ pub fn built_in_model_providers() -> HashMap<String, ModelProviderInfo> {
                 stream_max_retries: None,
                 stream_idle_timeout_ms: None,
             max_tokens: None,
-                max_tokens: None,
                 requires_openai_auth: true,
             },
         ),
@@ -384,7 +382,6 @@ pub fn create_oss_provider_with_base_url(base_url: &str) -> ModelProviderInfo {
         stream_max_retries: None,
         stream_idle_timeout_ms: None,
             max_tokens: None,
-        max_tokens: None,
         requires_openai_auth: false,
     }
 }