This includes the `jug pack` command for the NFS backend: running `jug pack` replaces small files by a bigger "packed file". This is particularly valuable if you have many many jobs that return small outputs (e.g., None, a filepath, a number, &c)
I have been using it internally myself for the past few months and it seemed robust and useful.