What is the problem the feature request solves?

The current microbenchmarks have been implemented in a variety of styles and there is a lot of code duplicated between each benchmark, such as running the benchmarks for "Spark", "Comet Scan", and "Comet Scan + Exec". Different benchmarks use different data sizes and different data generation approaches.

I would like to add microbenchmarks for all expressions that Comet supports, but I would like to refactor the existing benchmarks first so that we have a standard approach to use.

Describe the potential solution

No response

Additional context

No response