pub fn record_llm_usage( prompt_tokens: usize, completion_tokens: usize, total_tokens: usize, stop_reason: Option<&str>, )
Record LLM token usage on the current span