Improve LangChain memory implementation

- Swap to RunnableWithMemory
- Add verbosity flag

tests/test_llm.py

@@ -17,11 +17,12 @@ def test_chat_client(ollama_config, chat_client):
     if not is_ollama_available(ollama_config):
         pytest.skip("Local Ollama server is not available")
 
-    chat_client = chat_with_client(
+    response = chat_with_client(
         chat_client, "Tell me your name and introduce yourself briefly"
     )
-    response = chat_client.get_last_response_or_none()
 
+    response_from_history = chat_client.get_last_response_or_none().content
     assert response is not None
-    assert len(response.content) > 0
-    assert "gemma" in response.content.lower()
+    assert response == response_from_history
+    assert len(response) > 0
+    assert "gemma" in response.lower()