Check out the "archive the contents of this folder" link in file storage, it downloads a zip'd archive of the contents. Import into file storage isn't supported, though.
For .LRN as has been pointed out Ernie's working on exporting and importing in SCORM, which is an XML-based standard for interchange in that world.
For my own birdnotes site I wrote a bulk upload manager that was later adopted for the old ACS 3.x-based version of SloanSpace. We've mumbled about the fact that it would be very nice to have it for SloanSpaceV2 (.LRN), too, but thus far there's been no funding for it and I've been far too busy with other things to do it on my own time. I'd like to, though. It gave full control over bulk uploading of data, creating a log of rejected lines, allowing one to roll back uploads (useful if they fail for some reason), etc. You just implement a set of procs for a given file format for whatever dataset you're working with, and plug them into the bulk upload manager. In OpenACS 4.x one would do so via a service contract. This was designed for importing data from other sources (bird data made available by national organizations, etc.) not for interchange between ACS instances.
Interchanging data between ACS instances would be relatively difficult ... packages tend to work together, interweaving (say) forums content with attachments with, of course, the users and parties tables and notifications and ...
I don't see that custom file formats would be better than Oracle or Postgres's built-in ability to make consistent backups.
