v0.5.0 — Strategic Maze Elements

[0.5.0] strategic maze elements

Strategic Maze Elements: Complete overhaul adding 5 new strategic maze mechanics
Teleporters ('O'/'Q'): Single-use quick travel for strategic routing (+15 pts each, max 60)
Switches ('s'): Toggle gates ('S') for dynamic path changes (+20 pts each, max 80)
Movable Blocks ('B'): Push to create bridges/barriers (innovation bonus)
Bonus Exits ('F','G','H'): Optional objectives (+75 points each)
Conditional Doors ('X','Y','Z'): Complex unlock requirements (+30 pts each)
Enhanced Scoring System: Expanded from 5 to 8 scoring components
Strategic Innovation: Rewards creative use of strategic elements (0-100 pts)
Route Complexity: Multiple solution paths analysis (0-150 pts)
Bonus Objectives: Optional challenge completion (0-75 pts)
Reduced trap importance: Max 30 points vs unlimited before
Strategic Pathfinder: Advanced pathfinding algorithm handling stateful elements
Multi-dimensional state tracking (keys, switches, teleport usage)
Innovation analysis detecting creative strategic element usage
Timeout protection for complex maze solving
Enhanced Documentation: Updated prompt.md to guide models toward strategic thinking

Scoring Focus: Shifted from trap-spamming to strategic puzzle-solving
Typical scores increased from 200-400 to 800+ for quality strategic mazes
Quality over quantity approach for all maze elements
Maze Evaluation: Complete rewrite of strategic_evaluator.py
StrategicMaze class for intelligent element analysis
Enhanced pathfinding with stateful movement logic
Innovation detection and creative usage scoring
CLI Output: Updated to reflect new scoring components and strategic analysis
Strategic elements breakdown in results
Innovation details and unique strategies
Enhanced maze analysis with complexity metrics

Route Complexity Analysis: Resolved unpacking error in strategic element counting
Pathfinding State Management: Fixed teleporter and switch state tracking
Innovation Scoring: Corrected strategic element usage detection
Performance Issues: Optimized complex maze solving with proper timeout handling

AI Model Challenge: Higher intelligence ceiling for advanced models
Maze Variety: Eliminates repetitive trap-filling strategies
Strategic Depth: Multiple solution paths rewarded over single approaches
Benchmark Quality: True spatial reasoning assessment vs basic maze completion

Performance Validation: Sample strategic maze scored 871.69 points
Strategic Innovation: 70 points (teleporter + switch usage)
Route Complexity: 76 points (strategic element integration)
Quality Focus: Strategic placement prioritized over trap quantity
System Reliability: Robust handling of complex strategic element interactions
Backward Compatibility: Legacy maze evaluation maintained through grade_maze() wrapper