pub fn max_quantization_error(values: &[f64]) -> f64
Compute the maximum absolute quantization error across a slice of f64 values.