feat(scrape): 任意目录就地刮削,按路径识别作品,每次刮削一条记录(SCR-2)

- POST /files/scrape 改为对目录 / 视频文件刮削:已登记用作品,否则按路径识别(本地 + TMDB),
  认不准返回候选,选定后再提交;未登记直接用 TMDB 字段,addToLibrary 顺手建作品
- 作品详情重新刮削、入库后自动刮削改为对作品目录刮削,同一实现
- 迁移 0058_scrape_record 替代 media_scrape;GET /files/scrape-records
- 与 MP ScrapingChain 的文件布局差分(7 个场景,3 处登记偏离)

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

apps/server/test/file-manager/v2.test.ts

@@ -277,7 +277,7 @@ describe.skipIf(skip)("files v2(W9-H)", () => {
     expect(src.body.data.mediaEntityId).toBe(film);
   });
 
-  it("G5:批量刷新元数据按作品去重,未登记作品跳过", async () => {
+  it("G5:批量刷新元数据按作品去重(不要求有已入库文件)", async () => {
     const reg = await seedEntity(h.db, { kind: "movie", title: "Registered Film", year: 2018 });
     await mkdir(join(h.lib, "Registered Film (2018)"), { recursive: true });
     await writeFile(join(h.lib, "Registered Film (2018)", "a.mkv"), big);
@@ -299,8 +299,8 @@ describe.skipIf(skip)("files v2(W9-H)", () => {
       EXEC_KEY,
     );
     expect(r.status).toBe(200);
-    expect(r.body.data.submitted.map((s: Loose) => s.mediaEntityId)).toEqual([reg]);
-    expect(r.body.data.skipped).toEqual([{ ref: bare, reason: "not_registered" }]);
+    expect(r.body.data.submitted.map((s: Loose) => s.mediaEntityId)).toEqual([reg, bare]);
+    expect(r.body.data.skipped).toEqual([]);
   });
 
   it("G8:目录摘要 — 上一级、子目录是否有下级、视频/已登记计数", async () => {