We would like to run performance test for the time_series aggregation so that we are able to:

1. spot possible regressions with future changes
2. spot improvements we are planning to introduce (like pushing down filters from parent aggregations to child aggregations)

The typical use case is to have a (auto)date_histogram with a nested time_series aggregation or a terms aggregation with a nested time_series aggregation. At the moment we add the additional operations to the existing append-no-conflicts challenge to avoid reindexing data while running another challenge.