Add use cases in tests when loading repository

When running in production with a huge svn repository and the loading
broke, at restart from last known swh revision, a mismatch
happens. Triggering the history altered use case.

There was an initial problem with the underlying api used. Subvertpy did
not permit to inhibit the keyword expansion during the svn export, this
has been fixed upstream https://github.com/jelmer/subvertpy/pull/25,
packaged and deployed in our infra). Still, the problem persists and
for now i'm investigating it.

Those added use cases are a step towards finding why this happens.