Support calculating X annotations over Y datasets in the v2 api

[jgadling]

The frontend needs to display a statistic that calculates the number of datasets that have annotations for a given deposition. Right now the frontend is using this query:

query MyQuery {
  depositions {
    num_per_dataset: annotationsAggregate {
      aggregate {
        count
        groupBy {
          run {
            dataset {
              id
            }
          }
        }
      }
    }
  }
}

and uses aggregate.length to get the necessary, but we should determine if there's a better way to support this feature. Can/should we add more powerful aggregate functionality to the api in general? Should we make this a one-off special case?