Snapshot branch names were the url of sources. This url is used for
incremental loading: if an url is already existing in the last
snapshot, this url is not downloaded again.
Using the url as key for this cache has several drawbacks:
- if the upstream source is upgraded in place, the snapshot will reuse
the previous version of the upstream source.
- when the loader supports url mirrors, using the url as branch name will
be confusing (we won't know which url to use)
Moreover, this integrity attribute will be useful to get artifacts
from SWH once a metadata API will be available.