At the moment, the indexer, to work properly, needs to have at least some values inserted in the storage backend in the indexer_configuration table.
So far, that setup was done in the storage's sql procedure which is tedious to say the least (need to update storage, run sql script to the main db).

What would be good to have is self-contained indexer (so that we reduce commits in the swh-indexer repository, well possibly in swh-site if configuration changes).
Then, when they are deployed, the first action they do is push their configuration (at the moment, mostly what tool they use).

Then the usual implementation runs.