data about imports is stored in a slightly strange way

[Michael Hudson-Doyle]

Currently we store all data (aside from the branch) about an import in files named after the branch id in a single directory. It would make more sense to have a directory per import, stored broken down by id, as we do for branch data already.

We can implement this in a backwards compatible way by looking in the new location first, then the old location if nothing is found there, and always storing to the new location.