Created by: walshjon

This PR changes the accelerated resonance elastic scattering algorithm based on the double-integral change of variables (mu, v_t --> E_rel, E_t) so the E_rel and E_t are now both sampled directly and there is a single rejection criterion (that the resulting mu must be on [-1, 1]). Upscatter probabilities, scattering kernels, and k_eff values look very good using DBRC as a reference (the two methods should and do produce statistically indistinguishable results).

In preliminary testing with a pin cell and a single resonant isotope (U-238) I've observed this to be faster than DBRC, somewhat faster than ares, and within 1% of the standard constant cross section free gas model (cxs). More isotopes should compound this.