Integrate FSAL with librarian

[n0phx]

Since the metadata collection requires to store data in a similar manner as FSAL already does, it makes sense to use the same data source for both, which aside from saving disk space will reduce the number of queries too if fsal became part of librarian. Also, metadata collection would no longer require a separate indexing step, but fsal could schedule the metadata extraction while it performs the raw file indexing job.