Model XHR binary body and Content-Type echo responses plus six encoded document resources. Preserve raw request metadata and encoding bytes, avoid waiting for unused uploads, and discover only supported resource references. Match upstream cache and absent-body response headers.
Validation: 584 Python tests passed with 32 workers; 270 HTTP responses matched official wptserve for status, body, and 15 response fields. New regression tests fail on the unmodified main baseline.
Source-commit: 5389231812
ldm0
cecd00f3c1