v2.2 - Model Loading Optimization

Critical Fix: Models worden nu gecached en hergebruikt (4 models ipv 150+). Dit lost crashes op low-end devices op.

Main Fix:

• ✅ Model creation: 4 cached models hergebruikt met instances
• ✅ Memory: ~85% minder memory usage
• ✅ Performance: ~90% sneller laden
• ✅ Trees: Reduced 35→20, Fence: 80→48 posts

See CHANGELOG_v2.2.md voor details.