fn scale(bytes: u64, factor: f64) -> u64
bytes * factor, saturating and rounding down.
bytes * factor