feat: upgrading to opencv-4-12

[jeshernandez]

Summary Upgrading to OpenCV 4.12

• Image format improvements
• ARM performance enhancements Sources: https://opencv.org/blog/opencv-4-11-is-now-available/, https://opencv.org/blog/opencv-4-12-0-is-now-available/
• Potential performance gains on Raspberry Pi 4/5 (ARM Cortex-A72/A76) │
• KleidiCV provides optimized routines for image processing on ARM │
• Could improve ball detection speed and reduce frame processing latency │
• Better utilization of NEON SIMD instructions
• Better YOLO inference - Project uses ONNX YOLO models for ball detection
• ONNX improvements - More reliable ONNX model loading (readNetFromONNX)
• Interactive calibration tool enhancements Source: https://opencv.org/blog/opencv-4-12-0-is-now-available/
• Better camera calibration workflow (project has complex dual-camera calibration in gs_calibration.cpp)
• Could simplify calibration rig setup and testing
• More reliable ball detection under varying lighting conditions
• Fewer edge case failures in image processing
• Improved stability for edge cases in HoughCircles, CLAHE, and other algorithms