Simple adaptive quadrature for Ocaml

Following a C++ code I wrote doing the same thing I implement here Gauss-Kronrod adaptive quadrature in Ocaml.

Warnings

1. This implementation is slower than necessary.
2. This implementation is probably not robust.

Future Work

1. Would like to parameterize the type and not require it to be floating point.

2. Perhaps more work on module signature allowing a variety of integration methods, not just gauss-kronrod.

3. The adaptation algorithm here is very naive, as it is in my C++ code. This could be improved using more sophisticated logic and cut down on redundant function evaluations.