Allow multiprocessed scoring for `prime`

[bwhall61]

Right now the "priming" of the traverser (scoring the top layer nodes and initialize the priority queue) is single threaded. We should allow multiprocessing to make this faster for potentially slow scoring functions