clonamic-preprocessinglisted

# Clonamic Preprocessing Use only when normalization, material clarification, or a queue helps. Keep precise single items native. ## Runtime Use `scripts/preprocessing.py`; read [references/runtime-contract.json](references/runtime-contract.json) before queue persistence. - `normalize_text(text)` — normalize text without changing meaning. - `clarification_contract(text, missing_fields)` — ask only for caller-identified material gaps. - `enqueue`, `claim_next`, `record`, `queue_state` — operate a priority/FIFO queue at an explicit path. - `run_loop_auto(path, executor, enabled=True)` — drain queued items through a caller-supplied callable. The queue stores normalized text, priority, sequence, state, claim token, attempts, and result. Mutations use bounded locking and atomic replacement. Reclaimed stale work invalidates its old token. ## Explicit loop_auto Run only when the user request or accepted packet sets `loop_auto=true`; never infer it from size or queue length. The caller owns the executor and each item. Stop on executor error or step limit, preserve the failed item, and leave later items pending. Disabled means no calls or mutation. ## Result Return one `PreprocessingResult` with normalized input, questions, queued IDs, loop status, and runtime error. The caller owns execution and response.