feat(metadata): 作品进入关注时预热元数据缓存

- 新任务 prewarmMetadata(网络道):拉基础详情、富详情(含标志)、各季详情、选中的剧集组,
  海报 / 背景 / 标志原图进图片缓存;请求键与刷新元数据 / 刮削相同,之后直接命中且 pinned
- enqueueWatchedRefresh(建订阅、用户搜索 / 手动识别、提交下载)每次额外入队预热,幂等键按作品 + 日期

Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01Cxwgnn2pbzQ7rXv3rq7fei

apps/server/src/modules/sources-identity/tasks.ts

@@ -6,6 +6,7 @@ import { refreshSiteAccount } from "./service/account.ts";
 import { AUTO_ENRICH_TASK, runAutoEnrich } from "./service/enrich.ts";
 import { ingestFeed } from "./service/feeds.ts";
 import { refreshMetadata } from "./service/metadata.ts";
+import { prewarmMetadata } from "./service/metadata-prewarm.ts";
 import { resolveRelease } from "./service/resolve.ts";
 import { snapshotSite } from "./service/snapshot.ts";
 import { liveSearch } from "./service/workbench.ts";
@@ -77,6 +78,15 @@ export function registerSourcesTasks(registry: TaskRegistry, rt: () => SourcesRu
       handler: (t) =>
         t.step.run("refresh", () => refreshMetadata(rt(), t.input.mediaEntityId, { full: !!t.input.full })),
     }),
+    // 关注作品预热(统一元数据缓存):详情 / 各季 / 剧集组 / 图片先进缓存;网络道,走 TMDB 客户端限流
+    defineTask({
+      name: "prewarmMetadata",
+      version: 1,
+      lane: "network",
+      inputSchema: z.object({ mediaEntityId: id }),
+      retryPolicy: { maximumAttempts: 3, initialInterval: "5s", backoffCoefficient: 2, maximumInterval: "60s" },
+      handler: (t) => t.step.run("prewarm", () => prewarmMetadata(rt(), t.input.mediaEntityId)),
+    }),
     defineTask({
       name: "searchSubscription",
       version: 1,