Return internal information for future use

I'm currently doing a lot of cubature integrals, and it's quite slow due to the shear number of integrals. However, I suspect that the number of intervals and/or quadrature points at convergence are quite similar from one integral to another. Is there is an opportunity for speed up if this information can be passed from one cubature evaluation to another, so that the algorithm can start off from a better position?

Top work on this package btw!

The issue names no files, tests, or specific entry points. Start by tracing the cubature evaluation flow and how interval and quadrature-point convergence state is represented; define what state can be passed between evaluations and how reuse would be verified.

Written by the indexing model from the issue text.