Fixed possible race condition in renderer / collider batch threading Optimized all batch refreshes Cleanup