Currently, when fetching data with Repo.all/2, users have no way to obtain the count of
returned rows without either:
1. Calling Enum.count/1 on the results.
2. Running a separate Repo.aggregate.
However, database adapters already return the row count, Repo.all and Repo.all_by simply discards it by calling elem(1) on the {count, rows} tuple.