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.

apps/api/app/test_admin_ops.py

@@ -56,6 +56,8 @@ class AdminOpsTest(unittest.IsolatedAsyncioTestCase):
                 return {
                     "total_turns": 2,
                     "metered_turns": 1,
+                    "token_metered_turns": 1,
+                    "token_unmetered_turns": 0,
                     "tokens_in": 11,
                     "tokens_out": 13,
                     "cost_usd": 0.0042,
@@ -69,24 +71,33 @@ class AdminOpsTest(unittest.IsolatedAsyncioTestCase):
                         {
                             "day": "2026-06-28",
                             "turns": 1,
+                            "token_metered_turns": 1,
+                            "token_unmetered_turns": 0,
                             "tokens_in": 11,
                             "tokens_out": 13,
                             "cost_usd": 0.0042,
+                            "unpriced_turns": 0,
+                            "unpriced_tokens_in": 0,
+                            "unpriced_tokens_out": 0,
+                            "provider": "claude_cli",
+                            "model": "gateway-default",
                         }
                     ]
-                case.assertIn(
-                    "COALESCE(SUM(tokens_in), 0) + COALESCE(SUM(tokens_out), 0) DESC",
-                    query,
-                )
-                case.assertNotIn("tokens_in + tokens_out DESC", query)
+                case.assertIn("unpriced_tokens_in", query)
+                case.assertNotIn("LIMIT 12", query)
                 return [
                     {
                         "provider": "claude_cli",
                         "model": "gateway-default",
                         "turns": 1,
+                        "token_metered_turns": 1,
+                        "token_unmetered_turns": 0,
                         "tokens_in": 11,
                         "tokens_out": 13,
                         "cost_usd": 0.0042,
+                        "unpriced_turns": 0,
+                        "unpriced_tokens_in": 0,
+                        "unpriced_tokens_out": 0,
                     }
                 ]
 
@@ -101,13 +112,65 @@ class AdminOpsTest(unittest.IsolatedAsyncioTestCase):
         self.assertEqual(usage.source, "database")
         self.assertEqual(usage.total_turns, 2)
         self.assertEqual(usage.metered_turns, 1)
+        self.assertEqual(usage.token_metered_turns, 1)
+        self.assertEqual(usage.token_unmetered_turns, 0)
         self.assertEqual(usage.by_provider[0].provider, "claude_cli")
+        self.assertEqual(usage.by_provider[0].cost_basis, "provider_estimate")
+        self.assertEqual(usage.recorded_cost_usd, 0.0042)
+        self.assertEqual(usage.estimated_cost_usd, 0)
         self.assertTrue(usage.evaluator_cache.enabled)
         self.assertEqual(usage.evaluator_cache.requests, 4)
         self.assertEqual(usage.evaluator_cache.hit_rate, 0.75)
         self.assertEqual(usage.daily_cost[0].day, "2026-06-28")
         self.assertEqual(usage.daily_cost[0].cost_usd, 0.0042)
 
+    async def test_usage_from_database_backfills_agy_zero_cost_with_reference_rate(self) -> None:
+        class Conn:
+            def __init__(self) -> None:
+                self.fetch_calls = 0
+
+            async def fetchrow(self, query, *args, **kwargs):
+                return {
+                    "total_turns": 5,
+                    "metered_turns": 5,
+                    "token_metered_turns": 5,
+                    "token_unmetered_turns": 0,
+                    "tokens_in": 35_703,
+                    "tokens_out": 1_129,
+                    "cost_usd": 0,
+                }
+
+            async def fetch(self, query, *args, **kwargs):
+                self.fetch_calls += 1
+                row = {
+                    "provider": "agy_cli",
+                    "model": "gemini-3.6-flash-high",
+                    "turns": 5,
+                    "token_metered_turns": 5,
+                    "token_unmetered_turns": 0,
+                    "tokens_in": 35_703,
+                    "tokens_out": 1_129,
+                    "cost_usd": 0,
+                    "unpriced_turns": 5,
+                    "unpriced_tokens_in": 35_703,
+                    "unpriced_tokens_out": 1_129,
+                }
+                if self.fetch_calls == 2:
+                    return [{"day": "2026-07-31", **row}]
+                return [row]
+
+        with patch.object(admin_routes, "acquire", return_value=_Acquire(Conn())):
+            usage = await admin_routes._usage_from_database(window_days=30)
+
+        self.assertEqual(usage.cost_usd, 0.062022)
+        self.assertEqual(usage.recorded_cost_usd, 0)
+        self.assertEqual(usage.estimated_cost_usd, 0.062022)
+        self.assertEqual(usage.daily_cost[0].cost_usd, 0.062022)
+        breakdown = usage.by_provider[0]
+        self.assertEqual(breakdown.cost_usd, 0.062022)
+        self.assertEqual(breakdown.cost_basis, "reference_rate")
+        self.assertIn("Gemini 3.6 Flash", breakdown.rate_label or "")
+
     async def test_uptime_from_database_aggregates_health_samples(self) -> None:
         now = datetime.now(timezone.utc)