There is a lot of repetition among some google benchmarks in terms of the kinds of queries we want to test. It would be nice to have some sort of corpus to draw from that enumerates some interesting queries/properties (e.g. presence of indexes) so that each benchmark doesn't have to re-create queries that another benchmark has (and it seems like, when creating a benchmark, a lot more time is spent trying to come up with and writing the code to generate interesting queries than the benchmark itself). We had something similar for bonsai with bonsai_query_bm_fixture.h and maybe we can rename/extend that code to allow for more cases to draw from where appropriate.