Torch data loaders interfere with cleanvision parallelization

[sanjanag]

Need to set n_job=1 right now, otherwise the code runs super slow.

[jwmueller]

Easy solution for now:

in find_issues(): if the provided dataset was torch dataset and n_jobs = None, then just default n_jobs = 1

[jwmueller]

After fixing this, remember to remove the n_jobs = 1 argument currently passed for all the torch dataset examples/tutorials