Enhancing Precision of Average Treatment Effect Estimation in Causal Inference Through an Efficient Neighborhood Search Algorithm With Regard to Optimality