**Motivations:**
- total_sponsoring should be calculated from cache, not stored in tags
- Author searches should use cache first, not query Nostr directly
- All elements should be read/written from/to database
- Background sync should scan all notes with service='zapwall.fr' tag and update cache
- Sync should run at startup and resume on each page navigation
**Root causes:**
- total_sponsoring was stored in tags which required updates on every sponsoring payment
- Author queries were querying Nostr directly instead of using cache
- No background sync service to keep cache up to date with new notes
**Correctifs:**
- Removed totalSponsoring from AuthorTags interface and buildAuthorTags
- Modified getAuthorSponsoring to calculate from cache (sponsoring queries) instead of tags
- Modified parsePresentationEvent to set totalSponsoring to 0 (calculated on demand from cache)
- Modified fetchAuthorByHashId and fetchAuthorPresentationFromPool to use cache first and calculate totalSponsoring from cache
- Created platformSyncService that scans all notes with service='zapwall.fr' tag and caches them
- Modified _app.tsx to start continuous sync on mount and resume on page navigation
- All author presentations now calculate totalSponsoring from cache when loaded
**Evolutions:**
- Cache-first architecture: all queries check cache before querying Nostr
- Background sync service keeps cache up to date automatically
- totalSponsoring is always calculated from actual sponsoring data in cache
- Better performance: cache queries are faster than Nostr queries
- Non-blocking sync: background sync doesn't block UI
**Pages affectées:**
- lib/nostrTagSystemTypes.ts
- lib/nostrTagSystemBuild.ts
- lib/articlePublisherHelpersPresentation.ts
- lib/sponsoring.ts
- lib/authorQueries.ts
- lib/platformSync.ts (new)
- pages/_app.tsx
