fix(ci): pin Lahiri on CLI smoke fixtures after Raman default

Raman is the product default; these dignity and Dasha assertions were measured under Lahiri and must pass the ayanamsa explicitly.

Co-authored-by: Cursor <cursoragent@cursor.com>

docs/BUG_HISTORY.md

@@ -4971,6 +4971,22 @@
 - 复发自：BUG-328（当时为了露出聊天改成下半屏分栏，叠层低于输入框）
 - 修复版本：9670b661
 
+## BUG-331 | Raman 默认岁差后，Lahiri 校准的 CLI smoke 仍按旧盘面断言
+
+- 状态：resolved
+- 首次发现：2026-08-20
+- 最近更新：2026-08-20
+- 影响面：Gitea `backend-quality-gate` Python quick quality gate、`tests/test_cli_smoke.py`、`tests/golden/golden_cases.json` 只被 JSON 校验通过后进入同一套 pytest
+- 用户现象：质量门禁打印 `json ok tests/golden/golden_cases.json` 后继续跑 `audit_*` / `validate_bphs_invariants` / `pytest tests/test_cli_smoke.py ...`。JSON 本身有效；失败发生在 CLI smoke：Dasha 起始时刻、D9 Jupiter 尊严、Darakaraka 行星与 Navamsa 敌友标签对不上。
+- 触发条件：产品默认岁差改为 Raman 后，未给 Lahiri 校准样本显式传 `--ayanamsa lahiri`，再跑 quick quality gate。
+- 根因：这些断言锁的是旧 Lahiri 盘面（例如 `1952-03-19T20:24:47`、`NEECHA_BHANGA`、DK=`Sun`）。默认 Raman 后月亮分宿与分盘换升改变，断言失败。`golden_cases.json` 的 SAV=337 等不变量与岁差无关，JSON 校验和 golden runner 仍通过。
+- 修复：Lahiri 校准的 dignity / Dasha / DK 样本显式传 `--ayanamsa lahiri`。产品默认仍是 Raman；Raman 元数据测试继续显式传 `raman`。
+- 验证：`tests/test_cli_smoke.py` 中上述失败项；`tests/test_cli_smoke.py::test_full_reading_golden_cases_cover_user_ready_output`。
+- 防复发：对照样本必须显式写岁差名，不得依赖引擎默认；改 `DEFAULT_AYANAMSA_NAME` 后必须跑 `tests/test_cli_smoke.py` 与 quick quality gate 的 CORE pytest 列表。
+- 相关记录：无
+- 复发自：无
+- 修复版本：待发布
+
 ## BUG-329 | 生时纠正 Agent 回答在结算后一次性出现，推理中无法停止
 
 - 状态：resolved

tests/test_cli_smoke.py

@@ -24,6 +24,9 @@ BASE_BIRTH_ARGS = [
     "--lon", "116.4",
     "--tz", "8",
 ]
+# Lahiri-calibrated comparison fixtures. Product default is Raman; these
+# assertions lock Dasha/D9/DK labels that were measured under Lahiri.
+LAHIRI_ARGS = ["--ayanamsa", "lahiri"]
 
 
 def run_engine(*args: str) -> dict:
@@ -138,6 +141,7 @@ def test_chart_reports_richer_d1_dignity_labels_for_user_case() -> None:
         "--lat", "37.7749",
         "--lon", "-122.4194",
         "--tz", "8",
+        *LAHIRI_ARGS,
     )
 
     assert result["planets"]["Jupiter"]["sign"] == "Gemini"
@@ -227,6 +231,7 @@ def test_full_reading_accepts_second_and_preserves_birth_time_precision() -> Non
         "--lat", "37.7749",
         "--lon", "-122.4194",
         "--tz", "8",
+        *LAHIRI_ARGS,
         "--today", "2026-06-24",
         "--transit-date", "2026-06-24",
     )
@@ -699,7 +704,7 @@ def test_full_reading_exposes_sensitive_point_modules() -> None:
 
 
 def test_full_reading_uses_d9_context_for_navamsa_dignity() -> None:
-    result = run_engine("full-reading", *BASE_BIRTH_ARGS)
+    result = run_engine("full-reading", *BASE_BIRTH_ARGS, *LAHIRI_ARGS)
 
     d9_jupiter = result["modules"]["d9_navamsa_expanded"]["Jupiter"]
     assert d9_jupiter["sign"] == "Capricorn"
@@ -717,6 +722,7 @@ def test_full_reading_uses_d9_context_for_darakaraka_dignity() -> None:
         "--lat", "35.6895",
         "--lon", "139.6917",
         "--tz", "9",
+        *LAHIRI_ARGS,
     )
 
     darakaraka = result["modules"]["jaimini"]["darakaraka"]
@@ -736,6 +742,7 @@ def test_full_reading_uses_d9_context_for_vimsopaka_navamsa_dignity() -> None:
         "--lat", "35.6895",
         "--lon", "139.6917",
         "--tz", "9",
+        *LAHIRI_ARGS,
     )
 
     navamsa = result["modules"]["vimsopaka"]["Sun"]["varga_scores"]["Navamsa"]
@@ -743,7 +750,7 @@ def test_full_reading_uses_d9_context_for_vimsopaka_navamsa_dignity() -> None:
 
 
 def test_full_reading_maps_advanced_dignity_labels_into_vimsopaka_scores() -> None:
-    result = run_engine("full-reading", *BASE_BIRTH_ARGS)
+    result = run_engine("full-reading", *BASE_BIRTH_ARGS, *LAHIRI_ARGS)
 
     jupiter_d9 = result["modules"]["d9_navamsa_expanded"]["Jupiter"]
     assert jupiter_d9["dignity"] == "NEECHA_BHANGA"