as a follow-up to the thread on using Ollama for with_structured_output() instead of using OpenAI or Mistral, the ollama_functions.py needs to import from langchain_core.tools that imports from langchain_core.runnables.config for _set_config_context.