Function moe_expert_silu_down_shared 

pub fn moe_expert_silu_down_shared(
    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 with shared expert as extra blockIdx.y slot.

blockIdx.y < top_k: routed expert (pointer table + expert gate/up buffers). blockIdx.y == top_k: shared expert (direct pointers + sh_gate_in/up_in). Eliminates separate shared expert silu+down kernel launch.

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