List Smart Doc Extraction Runs

All available local eval runs, most recent first, for the run dropdown on the eval page. Each run id is a <results.json> file produced by one invocation of python -m backend.scripts.smart_doc_eval.