SlowLab: Benchmarking Autonomous AI Agents on a Year of Experiments They Cannot Undo | ArxivCSExplorer