Fix/optimization

[Myk72]

Summary

Refactors MetaMo core from application-specific research assistant code toward schema-driven, profile-based motivational state processing with diagnostics and pluggable policies.

Changes

• Added motivation schema support for goals, anti-goals, and modulators.
• Made MotivationalState and Action schema-aware and added action metadata.
• Introduced profile-based OpenPsi appraisal and MAGUS decision logic.
• Added goal-change calculation, MIC goal compatibility, reciprocal simulation results, and agent frame adaptation.
• Added diagnostics for bimonad transitions, law checks, safe-region status, drift, and blending.
• Refactored stability, coherence, and parallel merge behavior into default policy classes.
• Removed application-specific research assistant, papers, LLM, and core engine modules.