feat: add GitHub Analytics integration

[Empreiteiro]

Summary

• Add new GitHub Analytics data source type (github_analytics) distinct from the existing github_file source
• Backend: ask_github_analytics.py script that fetches 8 tables (repositories, pull_requests, issues, commits, releases, workflows, workflow_runs, contributors) via GitHub REST API with Link header pagination, loads into temporary SQLite, and uses LLM + SQL for Q&A
• Backend: github_analytics_router.py with test-connection, discover, and refresh-metadata endpoints
• Frontend: GithubAnalyticsSourceForm.tsx with token/owner/repo inputs, connection testing, and resource discovery panel
• Includes 5 report templates: PR Metrics, Issue Tracking, Commit Activity, CI/CD Health, Release Cadence
• Updated all integration points: ask.py routing, crud.py valid types, main.py router registration, AddSourceModal.tsx, apiClient.ts, and LanguageContext.tsx (EN/PT/ES)

Test plan

• Test connection with a valid GitHub PAT and public repo
• Verify resource discovery shows repo stats (stars, forks, issues, language)
• Connect a GitHub Analytics source and verify metadata refresh populates table schemas
• Ask natural language questions about PRs, issues, commits, workflows, and contributors
• Verify chart generation works with GitHub Analytics query results
• Test with private repos (requires PAT with repo scope)
• Verify i18n strings render correctly in PT and ES

Closes #70

🤖 Generated with Claude Code