initial benchmarks to try verifying O(2^k * n) for graph coloring

[andersenwall]

initial benchmarks complete for constant 2 bags with n nodes. verified n linear for false returns on coloring tests problem found with n worse than exponential for true returns on coloring tests