Add formal trait trade-off registry

[Spbd1]

Motivation

• Provide a canonical, machine-readable representation of context-dependent trait advantages and costs so simulation fitness models can reason about ecological trade-offs.
• Ensure no major trait is modeled as universally beneficial by encoding both positive and negative effects and capturing uncertainty via a confidence_level tag.
• Document the design and separate biology-inspired simulation assumptions from veterinary claims.

Description

• Add TraitEffect dataclass and TRAIT_TRADEOFFS registry in pishiegen/fitness/tradeoffs.py with the 12 required traits, each listing positive_effects, negative_effects, affected_contexts, biological_rationale, and confidence_level (high | moderate | speculative).
• Export the trade-off API from pishiegen.fitness by updating pishiegen/fitness/__init__.py to include TRAIT_TRADEOFFS, TraitEffect, and all_trait_effects.
• Add docs/trait_tradeoff_matrix.md documenting rules, the rationale summary for each trait, and the implementation link.
• Add tests in tests/test_trait_tradeoffs.py that assert the registry contains the required keys, every trait has at least one benefit and one cost, effect directions are explicit, and metadata fields (affected_contexts, biological_rationale, confidence_level) are present.

Testing

• Ran python -m pytest, all tests passed (20 passed), including tests/test_trait_tradeoffs.py which verifies the one-benefit/one-cost rule and metadata.
• Ran python -m compileall pishiegen to verify module byte-compilation succeeded.

---

Codex Task