Ewald: Implement prolate spheroidal wavefunction math rather than exp/erf

[blackwer]

We don't intend to use classical ewald summation techniques. We will be using the method outlined in https://arxiv.org/pdf/2505.09727v1 (known as ESP hah). The current Ewald code should be kept as a reference, and we should have a flag to swap between the methods. This will build off of #19. It's considerably more work, but important.