pub fn quantize_f32(value: f32, min: f32, max: f32, bits: u32) -> u32
Quantize a float to a fixed number of bits for bandwidth reduction.