Breaking The Curse Of Dimensionality In Electronic Structure Methods: Towards Optimal Utilization Of The Canonical Polyadic Decomposition