To talk to llamafile use any of the client libraries (https://github.com/dongri/openai-api-rs?tab=readme-ov-file, or https://github.com/tjardoo/openai-client), or async-openai, or try out codegen