Submitted by Hengjie Zhu 281 Trial by Trace: Benchmarking AI Agent Harnesses for Reliability and Failure Attribution TraceLite AI 19 6