pub fn dequant_fp8_to_bf16(fp8_bytes: &[u8], scale: f32, out: &mut [bf16])
Dequant FP8 (E4M3) bytes to BF16, applying a per-tensor scale.