feat(fetchers): WikipediaFetcher — clean article extraction via MediaWiki API

[chaliy]

What

Add a WikipediaFetcher that matches en.wikipedia.org/wiki/{title} URLs (and other language editions), returning clean article content via the MediaWiki API.

Why

Agents doing research and fact-checking frequently land on Wikipedia. The current DefaultFetcher returns the full page with edit links, references sections, navigation boxes, and other wiki-specific chrome. The MediaWiki API provides clean extract text and structured metadata.

Requirements

• Match: https://{lang}.wikipedia.org/wiki/{title} (all language editions)
• Fetch via API: https://{lang}.wikipedia.org/api/rest_v1/page/summary/{title} for summary
• Optionally fetch full content via: https://{lang}.wikipedia.org/api/rest_v1/page/html/{title}
• Return: title, extract/summary, infobox data (if parseable), key sections, categories
• Strip: edit links, reference numbers, navigation boxes, disambiguation notices
• Format field: "wikipedia"
• Support redirect resolution

Design Notes

• MediaWiki REST API is well-documented and has generous rate limits
• Summary endpoint returns a concise extract — often sufficient for agent needs
• Full HTML endpoint can be converted via existing html_to_markdown with wiki-specific cleanup
• Infobox extraction is complex — could be a stretch goal

Tier

2 — High-frequency agent need

[chaliy]

@bingxia2277 goal here is to make it avaible without extra APIs calls. You can use your Coding Agent to adopt fetchkit and benefit from fast rust implementaion.