Sparse Cholesky of Gram Matrix (SuiteSparse?)

[stevengj]

These matrices are big enough that the overhead of dynamic dispatch is probably negligible, so $ interpolation is just a good habit.