Long-report render fast path on by default (issue #121, BUG-1335)

Byte-identical to the legacy path on four fictional charts (PyJHora masked),
-19–26% end to end (median 23%). Short of the 25% written on the ticket; the
product owner chose to turn it on anyway. JYOTISH_REPORT_RENDER_FAST=0 restores
the old path.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoz1Qk3y9buDv1mkmTu

docs/BUG_HISTORY.md

@@ -18024,7 +18024,7 @@
 
 ## BUG-1335 | 个人长报告的中文出稿和溯源哈希重复劳动
 
-- 状态：investigating
+- 状态：resolved（2026-10-28 开关默认打开，产品选 A；待部署）
 - 首次发现 / 最近更新：2026-10-13 / 2026-10-14
 - 来源：issue #61。任务书 `docs/tasks/TASK-report-reference-perf-20261013.md`。
 - 现象：同一份个人长报告，中文正文要做多遍全文替换；大运补句的正则没有锚点，在长行上反复试；正文里根本没有的术语也会先编译再替换。溯源哈希对每一条「任意层的某个键」都把整包复制一遍。
@@ -18036,6 +18036,7 @@
 - 相关记录：BUG-693、BUG-694、BUG-576、BUG-1336。
 - 复发自：无。
 - 修复版本：分支 `codex/report-reference-perf-20261013`，默认未开通，尚未合入 staging。
+- 更新 2026-10-28（issue #121）：Linux 本机屏蔽 PyJHora、四张虚构盘各三次，开关开 / 关的报告逐字节相同（各字段 sha 一致），端到端快 19%～26%（中位 23%），单次溯源哈希约 0.55 s → 0.21 s。没达到 #121 写的 25%，产品选 A：照样默认打开（`RENDER_FAST_DEFAULT = True`），`JYOTISH_REPORT_RENDER_FAST=0` 可退回旧路径。原 35% 门槛由这次决定取代。验证：`tests/test_report_reference_perf.py`（快路径与旧路径逐字节相同）+ `scripts/research/report_reference_perf_ab.py --mask-pyjhora`。环境缺口：本机没装 PyJHora，不屏蔽 PyJHora 的那组待部署后在 staging 看报告生成用时。
 
 ## BUG-1336 | 同一张盘同一天再要长报告，每次都重算
 

scripts/report_perf_flags.py

@@ -8,12 +8,14 @@ from __future__ import annotations
 
 import os
 
-# Masked Windows A/B on 2026-10-14. Render shortcuts are byte-identical and use
-# less memory, but the end-to-end drop was 20–23%, short of the 35% gate, so
-# the fast path stays off. Cache hits matched the miss bytes and returned in
-# about 0.06s, under the 1.5s gate. A missing or unknown GITHUB_SHA still
-# disables the cache at the call site.
-RENDER_FAST_DEFAULT = False
+# Masked Windows A/B on 2026-10-14: render shortcuts byte-identical, less memory,
+# end-to-end -20–23%, short of the 35% gate then set, so the path stayed off.
+# Linux A/B on 2026-10-28 (four fictional charts, PyJHora masked): byte-identical
+# again, -19–26% (median 23%). The product owner chose to turn it on anyway
+# (issue #121, option A); JYOTISH_REPORT_RENDER_FAST=0 restores the old path.
+# Cache hits matched the miss bytes and returned in about 0.06s, under the 1.5s
+# gate. A missing or unknown GITHUB_SHA still disables the cache at the call site.
+RENDER_FAST_DEFAULT = True
 REFERENCE_CACHE_DEFAULT = True
 
 _ON = {"1", "true", "yes", "on"}