pub fn sample_float(low: f64, high: f64, scale: Scale, t: f64) -> f64
Sample a float from [low, high] given t in [0, 1), respecting scale.