How a page is built

Scrape — npm run scrape fetches live HTML + /_next/static assets into public/ (same paths, zero rewrites), splits header/body/footer into components/exact/, extracts giant inline base64 PNGs to public/images/, saves telda's stylesheet as public/telda-core.css.
Translate — node scripts/translate-ar.mjs builds the AR landing from the EN home files (61-string EN→AR map, reports leftovers).
Render — components/TeldaPage.tsx injects markup and boots behaviours.