Currently, there is a potential issue where cores^2 worker threads may be spun up to run parallel collection validation.  Explore the possibility of using a shared pool for workers on the inner loop across record store slices.