run_without_ray_tune fails for continual learning

It is noted here: https://github.com/numenta/nupic.research/issues/637 that we should not use the flag run_without_ray_tune but is there a way this can be fixed? I'm using a slurm server that is not compatible with ray. 