Data-Driven Approximation Of Transfer Operators: Dmd, Perron–Frobenius, And Statistical Learning In Wasserstein Space