Reproducibility in Matrix and Tensor Decompositions: Focus on Model Match, Interpretability, and Uniqueness