feat(identity): content unit data hygiene — specials, provisional promotion, duplicate merge (DH-1)

- metadata.includeSpecials (default true): full refresh and the language
  backfill fetch TMDB season 0 (specials) when present, like MP; season 0
  stays out of the episode-total ledger
- refreshMetadata promotes provisional season/episode units that match the
  TMDB season/episode layout to confirmed (source=tmdb), keeping ids and all
  references; non-matching units stay provisional
- content_unit.source (migration 0053): release / file / tmdb / user
- duplicate units (same entity/kind/season/episode/content type/segment) are
  merged as a refresh step and via `scripts/identity-cleanup.ts
  --merge-duplicates` (dry-run by default); keeper = most referenced, then
  confirmed, then earliest; references auto-discovered via FKs and repointed
- ensureUnits takes a per-unit advisory xact lock to stop concurrent lazy
  creation from producing duplicates
- runbooks updated; user DB read-only dry-run: 征途 S01E09/E10 4 units each
  -> 1 each (6 removed), not applied

Co-Authored-By: Claude Opus 5.5 <[email protected]>

docs/plan/00-first-version-plan.md

@@ -173,3 +173,4 @@ NFO / 图片投影、身份巡检与挂载根因聚合、HTTP 失败自动切浏
 - **浏览器会话重启自愈(BR-3,用户 2026-09-29 反馈"标签页不存在或已关闭")**:camofox 重启后旧 tab 404 → 服务端自动重建上下文、注入 Cookie、回到原 URL 并重试;`POST /browser/sessions/{id}/reconnect`;前端自动重连一次并提示;容器实例变化时全部会话标 needs_reconnect。
 - **VER-3 真机核对(2026-09-29)**:通过——登录态注入、搜索两阶段、待处理 6、Agent 抽屉、快照字段/时区/手机折叠;问题——严重:RSS 标题「[电视剧]」前缀被解析为中文名导致 15 条误映射到《东北有个周东北》(ID-2 修:分类前缀、别名守卫、快照更新重解析、错映射撤销脚本);中:RSS 条目未规范化(分类/大小/标题)、详情页只抓到 MediaInfo/豆瓣/标签(HH-2 修);低:请求计数不持久、「12 Memories」误判 E12、会话页乱码字符、原站视图默认显示「Agent」。
 - **数据治理后续(FIX-6 之后)**:全量回刷覆盖季 0(特别篇);TMDB 已确认的季集单元由临时转正式(沙丘:预言 仍为临时);同作品同季同集重复临时单元合并(征途 E09/E10 各 3 条);清理脚本修正后对用户库 dry-run 为 0 删除,未执行 apply。
+  - DH-1 已做(分支 dh1):`metadata.includeSpecials`(缺省开)full 刷新 / 回刷含季 0;刷新时对上 TMDB 季集结构的临时单元转正式(`content_unit.source`,迁移 0053);重复单元合并(refresh 一步 + `identity-cleanup.ts --merge-duplicates`),懒建加咨询锁防竞态。用户库只读 dry-run:征途 S01E09/E10 各 4 条 → 各保留 1 条,共删 6 条,未 apply。见 runbook metadata.md / identity.md。