レコードをメール: Forecasting Dynamic Correlations Between Carbon, Energy, and Stock Markets Using a BOHB-Optimized Multivariable Graph Neural Network