Implement benchmarks

[pasabanov]

Problem or Motivation

Currently, the library does not include benchmarks, which prevents comparing the performance of different versions of functions as well as different optimization levels.

Solution or Suggestion

It is proposed to create a benches folder in the project root and implement benchmarks there.

Framework Choosing

Below is a comparison table of benchmark frameworks:

Framework	Dependencies	Header-only	User-defined Metrics Support	Development/Support
Without framework	✅ None	—	⚠ Yes, but must be implemented manually	⚠ Depends on the developer
Catch2	✅ Common for tests and benchmarks	❌ No	❌ No	✅ Actively maintained
Google Benchmark	⚠ 1	❌ No	✅ Yes	✅ Actively maintained (by a large IT company)
Celero	⚠ 1	❌ No	⚠ Yes, but with complex syntax	✅ Actively maintained
criterion	⚠ 1	✅ Yes	❌ No	❌ Not maintained (last update 2020.11.03)
picobench	⚠ 1	✅ Yes	❌ No	⚠ Not actively maintained (last update 2024.03.06)

Since support for user-defined metrics is critical, the Google Benchmark framework was chosen.

Tasks

1. • Create a statistics module (namespace) with functions for standard deviations that are intended to be used for user-defined metrics. Introduced alfi::util::stat (statistics) namespace #50
2. • Implement benchmarks using the Google Benchmark framework and the statistics module in the benches folder.