Add MediaWiki diff enrichment worker service

[artvandelay]

Motivation

• Add a background worker to consume stored events and enrich them with MediaWiki diffs and page metadata for downstream analysis and political scoring.
• Extract structured added/removed text and surrounding context from MediaWiki HTML diffs for reliable artifact storage.
• Cache page-level metadata (categories and Wikidata QID) with TTL to avoid repeated API calls.
• Handle API rate limits and transient errors with retry/backoff to improve robustness.

Description

• Add a new worker EnrichWorker in services/enrich/worker.py that fetches pending events, obtains diffs with action=compare, parses HTML, persists diff records, and triggers metadata fetching via MetadataFetcher.
• Implement parse_diff_html in services/enrich/diff_parser.py which converts MediaWiki HTML diffs into DiffFragment objects containing added_text, removed_text, and context.
• Provide SQLite-backed storage DiffStorage in services/enrich/storage.py that creates events, diffs, and metadata_cache tables and exposes methods to fetch/mark events, insert diffs, and upsert/get cached metadata.
• Add HttpClient with retry and exponential backoff behavior in services/enrich/http_client.py and MetadataFetcher in services/enrich/metadata.py that fetches categories and wikibase_item QID and caches results with a TTL.

Testing

• Run the diff parser unit test with python -m unittest tests/test_diff_parser.py which executed the parse_diff_html parsing scenarios and succeeded.
• The test suite ran 1 unit test (tests/test_diff_parser.py) and passed (OK).

---

Codex Task