It would be useful to support OpenMC's uniform_source_sampling setting in the NWL calculation, to better sample the source mesh. This will also require that the calculation takes particle weights into account, since particle birth weights are affected by uniform_source_sampling. This will need to wait until OpenMC's develop branch is merged into master upon a new release, since unstructured meshes are not currently supported for openmc.MeshSource, and openmc.stats.MeshSpatial, which we currently use, is not affected by the uniform_source_sampling setting.