feat(api): backfill article markdown storage

[sommio]

Summary

Store extracted article markdown during feed ingestion so each article
can be consumed as readable text without extra fetches and reparsing.

What Changed

1. Add contentMarkdown and contentExtractedAt to Article via
   migration and Prisma model updates.
2. Add article-content module with repository, extraction service,
   and dedicated controller for content operations.
3. Extract readable text from article HTML with Readability and
   Turndown, normalize markdown quality, and skip unreadable or
   empty outputs.
4. Extend feed ingestion to trigger backfill after article persistence,
   with per-feed and per-article timeout budgets.
5. Add POST /article-content/:id/retry for forced reprocessing and add
   unit/e2e coverage for extraction, timeout handling, and retry flows.
6. Update docs and runbooks in English and Chinese, plus API/PR
   README adjustments and dependency lock updates.

Example Flow

flowchart TB
A[Ingest feed] --> B[Parse and normalize items]
B --> C[Persist article rows]
C --> D[tryPersistArticleContent]
D --> E{Markdown exists?}
E -->|yes| F[Skip]
E -->|no| G[Fetch article with timeout]
G --> H[Readability + Turndown]
H --> I[Save contentMarkdown and extractedAt]

Loading

---