Fix writing assist: disable thinking mode, drop stuck-buffer 409

- stream_chat: add think=False parameter passed through to Ollama payload.
  qwen3 models have thinking enabled by default; without this flag the model
  spends minutes generating internal thinking tokens that stream_chat silently
  discards, leaving the frontend spinner blank until the SSE connection times
  out and the widget disappears.

- create_assist_buffer: orphan (overwrite) a still-running buffer instead of
  raising. The old asyncio task holds a direct reference and completes
  harmlessly against the stale buffer. New requests always win.

- assist_route: remove the 409 guard that blocked new requests when a previous
  generation got stuck. create_assist_buffer now handles this transparently.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

src/fabledassistant/services/generation_buffer.py

@@ -100,7 +100,9 @@ def create_assist_buffer(user_id: int) -> GenerationBuffer:
     key = f"assist:{user_id}"
     existing = _buffers.get(key)
     if existing and existing.state == GenerationState.RUNNING:
-        raise RuntimeError(f"Assist generation already running for user {user_id}")
+        # Orphan the old buffer — the background task holds a direct reference
+        # and will complete against it harmlessly. A new request always wins.
+        logger.warning("Assist generation still running for user %d; orphaning old buffer", user_id)
     buf = GenerationBuffer(conversation_id=0, assistant_message_id=0)
     _buffers[key] = buf
     return buf