perf: reuse embedding server for query embed

examples/main_cli_example.py

@@ -1,7 +1,3 @@
-import faulthandler
-
-faulthandler.enable()
-
 import argparse
 from llama_index.core import SimpleDirectoryReader, Settings
 from llama_index.core.node_parser import SentenceSplitter
@@ -62,7 +58,7 @@ async def main(args):
 
     print(f"\n[PHASE 2] Starting Leann chat session...")
 
-    llm_config = {"type": "hf", "model": "Qwen/Qwen3-8B"}
+    llm_config = {"type": "hf", "model": "Qwen/Qwen3-4B"}
 
     chat = LeannChat(index_path=INDEX_PATH, llm_config=llm_config)