Function moe_expert_silu_down_shared_batch3 

pub fn moe_expert_silu_down_shared_batch3(
    gpu: &dyn GpuBackend,
    kernel: KernelHandle,
    gate_out: DevicePtr,
    up_out: DevicePtr,
    packed_ptrs: DevicePtr,
    scale_ptrs: DevicePtr,
    scale2_vals: DevicePtr,
    output: DevicePtr,
    expert_indices: DevicePtr,
    sh_gate_in: DevicePtr,
    sh_up_in: DevicePtr,
    sh_down: &QuantizedWeight,
    sh_down_out: DevicePtr,
    n: u32,
    k: u32,
    top_k: u32,
    stream: u64,
) -> Result<()>

Fused SiLU+down expert GEMV for K=3 tokens with shared expert.

Grid: (ceil(N/8), 3*(top_k+1), 1) Block: (128, 1, 1)