Operator: HardTanh-Affine-Gate (Fused CUDA Kernel)

Goal
- Fuse affine, HardTanh activation, sigmoid gate, and multiply to reduce memory traffic and improve throughput (≥1.30x).

Inputs/Outputs
- Input `x`: [B, D], float32
- Parameters `scale`, `bias`: [D], float32
- Scalars `alpha`, `beta`: float32
- Output `y`: [B, D], float32

Definition
- z = x * scale + bias
- m = clamp(z, -1, 1)
- g = sigmoid(alpha * m + beta)
- y = x * g
