Option for non-random queries

[NathanNeff]

Perhaps an option for non-random queries would be handy?

For example, I have a dataset that's text and the same dataset that's
stored as Parquet and I'd like to do a blanket comparison by running a set of queries
on both datasets.

Of course, a much more scientific approach for benchmarking should be used,
but at least if we know that Query #1 ran X times and Query #2 ran X times also, then we'd
be more confident in results over a set of time, rather than queries being executed at random.