Create benchmark to compare multi-planner query selection against cost-based-ranker using sampling CE and evaluate the performance difference between the two alternatives over a set of queries..

The outcome of this ticket should be an easy to repeat microbenchmark evaluating the sampling CE implementation and comparing it against the multi-planner alternatives.
The implementation should use the already pre-existing infrastructure for data generation and comparison as already used for the sampling and histogram ce performance benchmarks and existing CBR tests comparing with MultiPlanner