feat: Implement LJPW Dynamic System Model with predictive and corrective capabilities

Major features:
- Dynamic simulation using differential equations
- Predictive trajectory analysis (converging, diverging, oscillating, collapsing)
- Intervention planning with actionable recommendations
- Stability analysis with real-time concern detection
- Context-specific recommendations (cybersecurity, code, organization)

Core modules added:
- guardian/core/dynamics.py: DynamicLJPW simulator with Euler integration
- guardian/core/intervention.py: InterventionEngine for corrective action planning
- tests/test_dynamics.py: Comprehensive test suite (34 tests)
- tests/test_intervention.py: Intervention testing (35 tests)

CLI commands added:
- `guardian simulate`: Simulate LJPW system evolution over time
- `guardian intervene`: Generate intervention plans for system correction

Features:
- Predefined scenarios (love_deficit, wisdom_intervention, reckless_power, balanced_growth)
- Natural Equilibrium as stable attractor (0.618, 0.414, 0.718, 0.693)
- Love force multiplication effect
- Reckless Power detection (high P + low W erodes Justice)
- Wisdom-based intervention prioritization
- Timeline estimation for corrections

Documentation:
- docs/DYNAMIC_SYSTEM_GUIDE.md: Complete user guide with examples

All tests pass (140 passed, 2 xfailed, 69 new tests added)
Zero dependencies added (pure Python implementation using math module)

This transforms Guardian from descriptive (analyzing current state) to
predictive and prescriptive (forecasting evolution and recommending interventions).

Author: claude