v0.3.0

Added

• Predator-Prey Optimisation Heuristic: Implemented a new optimisation phase in Engine.multiply_all_vectors that
  runs before Jaccard similarity clustering. This "predator-prey" logic uses small StateVector instances (e.g., from
  evidence) to significantly reduce the size of larger "prey" vectors, dramatically improving compilation performance on
  many rule sets.

• Optimisation Parameter API: Added an .opt_config property to the Engine class. This allows advanced users to
  tune the optimisation heuristics (e.g. max_predator_size, max_cluster_size) by modifying internal attributes like
  _opt_max_predator_size on an engine instance.

Changed

• Refactored the Engine.multiply_all_vectors method, splitting the core optimization logic into smaller, internal
  static methods (_update_multiplication_state) to improve readability and maintainability.