G0~G8 성과·동맹 측정 OS 작업 일괄 고정

8월 7일까지 워킹트리에만 남아 있던 미커밋 작업을 커밋한다. 여러 사본
폴더(worktree·clone)에 흩어져 있던 중간 스냅샷을 정리하기 전에 원본을
git 이력으로 고정하는 것이 목적이다.

- contracts/routes/services: measurement, outcome_trajectory, rupture_repair,
  deliberate_practice, calibration_transfer, supervision_research,
  multimodal_alliance, continuous_improvement 계열 신규 모듈과 테스트
- infra/db/init: 07~16 마이그레이션(측정 기반~calibration transfer 실행)
- apps/web: 세션 리뷰 카드·관리 화면·E2E 스펙 추가
- docs/ops: G0~G8 라이브 통합·배포·롤백 증거 문서와 evidence JSON/PNG
- scripts: smoke·ledger·릴리스 에이전트·NAS 프리뷰 운영 스크립트

engine.public 로그 .bak과 apps/web/test-results 산출물은 커밋에서 제외했다.

scripts/compare-alliance-calibration.py

@@ -0,0 +1,86 @@
+#!/usr/bin/env python3
+"""Compare two versioned Alliance calibration snapshots against the G0 gold pack."""
+
+from __future__ import annotations
+
+import argparse
+import json
+import sys
+from pathlib import Path
+from typing import Any
+
+
+REPO_ROOT = Path(__file__).resolve().parents[1]
+API_ROOT = REPO_ROOT / "apps" / "api"
+if str(API_ROOT) not in sys.path:
+    sys.path.insert(0, str(API_ROOT))
+
+from app.contracts.measurement import BenchmarkCase  # noqa: E402
+from app.services.alliance_calibration import (  # noqa: E402
+    AllianceCalibrationSnapshot,
+    compare_alliance_snapshots,
+    gold_reference_snapshot,
+    render_alliance_comparison_markdown,
+)
+
+
+def _read_json(path: Path) -> Any:
+    return json.loads(path.read_text(encoding="utf-8"))
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--baseline", type=Path)
+    parser.add_argument("--gold-baseline", action="store_true")
+    parser.add_argument("--candidate", type=Path, required=True)
+    parser.add_argument(
+        "--benchmark",
+        type=Path,
+        default=API_ROOT / "app" / "data" / "outcome_alliance_benchmark_g0.v1.json",
+    )
+    parser.add_argument("--out-json", type=Path, required=True)
+    parser.add_argument("--out-md", type=Path, required=True)
+    parser.add_argument("--min-direction-accuracy", type=float, default=0.0)
+    args = parser.parse_args()
+
+    benchmark_payload = _read_json(args.benchmark)
+    cases = tuple(
+        BenchmarkCase.model_validate(item) for item in benchmark_payload["cases"]
+    )
+    if args.gold_baseline == (args.baseline is not None):
+        parser.error("choose exactly one of --baseline or --gold-baseline")
+    baseline = (
+        gold_reference_snapshot(cases)
+        if args.gold_baseline
+        else AllianceCalibrationSnapshot.model_validate(_read_json(args.baseline))
+    )
+    candidate = AllianceCalibrationSnapshot.model_validate(_read_json(args.candidate))
+    report = compare_alliance_snapshots(baseline, candidate, cases)
+    args.out_json.parent.mkdir(parents=True, exist_ok=True)
+    args.out_md.parent.mkdir(parents=True, exist_ok=True)
+    args.out_json.write_text(
+        json.dumps(report, ensure_ascii=False, indent=2, default=str) + "\n",
+        encoding="utf-8",
+    )
+    args.out_md.write_text(
+        render_alliance_comparison_markdown(report),
+        encoding="utf-8",
+    )
+    accuracy = float(report["candidate"]["direction_accuracy"])
+    print(
+        json.dumps(
+            {
+                "candidate_direction_accuracy": accuracy,
+                "candidate_evidence_recall": report["candidate"]["evidence_recall"],
+                "candidate_missing_or_failed": report["candidate"]["missing_or_failed_count"],
+                "mean_absolute_score_delta": report["drift"]["mean_absolute_score_delta"],
+                "report": str(args.out_md),
+            },
+            ensure_ascii=False,
+        )
+    )
+    return 0 if accuracy >= args.min_direction_accuracy else 2
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())