Optimizing Digital Marketing Revenue Forecasting Using an XGBoost–Dipper Throated Optimization Hybrid Model