Could somebody point me to some actors that cache data to make future workflow runs more efficient? I know that the provenance framework being worked on would require something like this, but I'm wondering if any actor has done this on it's own? I'm just looking for ideas...