Function shard_dense_1d_bf16 

pub fn shard_dense_1d_bf16(
    src: DevicePtr,
    dim: usize,
    tp_rank: usize,
    tp_size: usize,
    gpu: &dyn GpuBackend,
) -> Result<(DevicePtr, usize)>

Shard a 1D BF16 vector [dim] (e.g. q_norm_full, gate_proj bias) on dim 0. Used for per-output vectors that pair with column-parallel GEMMs.