Extension of the Dip-test Repertoire -- Efficient and Differentiable p-value Calculation for Clustering