Steady State GA: Lifecycle Hooks & Tests

[francastagna]

PR Summary

We implemented and validated the Steady-State GA behavior and added comprehensive tests:

• Steady-State GA now emits generation/step lifecycle events and uses score with frozen targets when deciding replacements.
• New deterministic tests verify steady-state replacement logic and edge cases.

---

Main Changes

• Steady-State GA (SteadyStateGeneticAlgorithm):
  • Invokes lifecycle hooks: beginGeneration, endGeneration, beginStep, endStep.
  • Freezes targets each iteration (frozenTargets = archive.notCoveredTargets()).
  • Uses score (respects frozenTargets) for replacement:
    • Replace the two selected parents with the two offspring only if max(score(o1), score(o2)) > max(score(p1), score(p2)).

Tests

• Steady-State GA tests added/updated (SteadyStateGeneticAlgorithmTest.kt):
  • Finds optimum on OneMax.
  • Edge cases:
    • xoverProbability = 0.0 → no crossover; two mutations; population size preserved.
    • fixedRateMutation = 0.0 → crossover occurs; zero mutations; population size preserved.
  • Replacement behavior (deterministic via FixedSelectionStrategy) → next generation is created as expected.