This PR fixes two minor issues in the simulation used to generate the graphics for BIP-0119.
- Use the same random seed across runs of the simulation; this ensures that it's an apples-to-apples comparison
- Fix an issue where a distribution was poisson and not exponential as it should have been. Poisson and exponential have the same mean as used, but exponential has mean squared variance, resulting in the curves for Congestion Controlled transactions being too "smooth".