fix(chat): complete streaming for reasoning, tools, and tool calls

- Forward reasoning_content and non-stream reasoning to on_token in
  OpenAI chat completions; Gemini thought parts to on_token.
- Stream bubble: always render session.tool panels; format tool_use_call
  payloads and title as call · tool_name.
- Emit tool_use_call via on_tool_ui before tool execution so WS shows
  invoke cards before tool_use_result.

Co-authored-by: Cursor <cursoragent@cursor.com>

platform/llm/transports/google_gemini_sse.py

@@ -224,6 +224,8 @@ class GoogleGeminiChatModel(ChatModel):
                                 txt = str(p.get("text") or "")
                                 if txt:
                                     thinking_parts.append(txt)
+                                    if on_token:
+                                        on_token(txt)
                                 sig = p.get("thoughtSignature")
                                 if isinstance(sig, str) and sig:
                                     thinking_signature = sig

platform/llm/transports/openai_chat_completions.py

@@ -551,8 +551,11 @@ class OpenAIChatModel(ChatModel):
         reasoning_parts = getattr(msg, "reasoning_content", None) or ""
         reasoning_text = str(reasoning_parts).strip() if reasoning_parts else ""
         content = msg.content or ""
-        if on_token and content:
-            on_token(content)
+        if on_token:
+            if reasoning_text:
+                on_token(reasoning_text)
+            if content:
+                on_token(content)
 
         tool_calls: list[LLMToolCall] = []
         if msg.tool_calls:
@@ -620,6 +623,8 @@ class OpenAIChatModel(ChatModel):
                 rc = getattr(delta, "reasoning_content", None) or ""
                 if rc:
                     reasoning_parts.append(rc)
+                    if on_token:
+                        on_token(rc)
                 if delta.tool_calls:
                     for tc in delta.tool_calls:
                         idx = int(tc.index)