Function quantize_per_channel 

pub fn quantize_per_channel(
    tensor: &Tensor,
    channel_axis: usize,
) -> Result<PerChannelQuantResult, ModelError>