feat: add injectable Python behavior strategies for Organism

[YJack0000]

Summary

• Add ReactionStrategy and InteractionStrategy as injectable std::function types on Organism
• Python users can now fully customize how organisms behave without touching C++ code
• Strategies are inherited by offspring via reproduce()
• Default behavior unchanged when no custom strategy is set

Usage Example (Python)

from simevopy import Organism, Genes, Food

# Custom reaction: flee from everything
def flee_strategy(organism, nearby_objects):
    if not nearby_objects:
        return (0.0, 0.0)
    my_pos = organism.get_position()
    obj_pos = nearby_objects[0].get_position()
    return (my_pos[0] - obj_pos[0], my_pos[1] - obj_pos[1])

# Custom interaction: pacifist (only eat food, never kill)
def pacifist_strategy(organism, nearby_objects):
    for obj in nearby_objects:
        if hasattr(obj, 'can_be_eaten') and obj.can_be_eaten():
            organism.add_life_span(500)

org = Organism(Genes(chr(40) * 4))
org.set_reaction_strategy(flee_strategy)
org.set_interaction_strategy(pacifist_strategy)

New Python API

Method	Description
organism.set_reaction_strategy(fn)	Set custom reaction to nearby objects. fn(organism, objects) -> (dx, dy)
organism.set_interaction_strategy(fn)	Set custom interaction logic. fn(organism, objects) -> None
organism.add_life_span(amount)	Modify lifespan from custom strategies

Discussion Points

• handleInteractions 和 handleReactions 改為 single-threaded 以避免 Python callback 的 GIL deadlock。未來如果要恢復多線程，需要在 C++ 端做 GIL release/acquire 管理
• 目前 nearby_objects 包含所有 EnvironmentObject（Food + Organism），Python 端需要用 hasattr 或 isinstance 判斷類型。考慮是否要提供 typed accessor 來簡化
• Strategies 是 per-organism 的。如果想要 per-species 共用，可以在 Python 端用同一個 function reference

Test plan

• All 3 existing Python tests pass
• Custom Python strategy smoke test passes
• Test strategy inheritance through reproduce()
• Test mixed default/custom organisms in same environment

🤖 Generated with Claude Code