Add an instruction-level performance microbenchmark (or microbenchmarks) for a simple aggregation pipeline. The aggregation stages chosen should align with the top queries sorted by Atlas MMR so the benchmark reflects customer-representative patterns (see top_ten_queries_locust HVW).

Before implementation, check existing microbenchmark coverage to avoid duplication.

The benchmark will be added to PR checks and/or a triaged context in separate work within the project.

This is a good candidate to also measure wall-clock latency if instruction count does not trend with end-to-end workloads.

Acceptance Criteria:

• A simple aggregation microbenchmark (or microbenchmarks) is implemented and running on the waterfall.
• The benchmark should meet the following standards:
  • Trends with our HVWs
  • Low noise
  • Short