Mistake: counting URLs but ignoring the 50MB byte ceiling

Mistake: counting URLs but ignoring the 50MB byte ceiling

The limit is two numbers, not one: 50,000 URLs and 50MB uncompressed per file. Long URLs with image/video extensions blow the byte limit well before 50k rows.

Fix checklist:
— ✅ Each file under 50,000 URLs
— ✅ Each file under 50MB uncompressed (measure before gzip)
— Estimate: plain URL sets rarely hit 50MB, but image sitemaps with 5+ images per URL can cross it at ~10,000 rows
— Split on whichever limit you hit first
— Gzip is allowed and recommended, but the 50MB cap applies to the decompressed size

Definition of done: a monitoring check that fails the build if any sitemap file exceeds 49,000 URLs or 48MB uncompressed. Leave headroom; do not ship at exactly 50k.