The permuter should be able to work off of improvements it has already found, deepening the search space. Ideally this would work when running multithreaded.

Each Permuter would have an immutable starting point, and main.py or something else would create new Permuters when improvements to a base are found. We also need to worry about caching results between Permuters to avoid storing the same improvements multiple times. (simon to fill in details)

One pitfall of this approach is that it's sure to generate tons of noisy mutations to the original code, especially with each new layer of findings. To address that, we should try to eliminate meaningless code mutations (#155).