contribution: gate cnn data generator with template

[mikesmalbroek]

This pull request adds a script to generate synthetic gate images for training purposes. The goal is to quickly create a dataset with varying gate positions, orientations, and visual conditions without relying entirely on manually collected data.

The script applies several simple transformations such as perspective distortion, random placement, noise, and motion blur to make the images less uniform and more representative of real flight conditions. Along with the images, a labels.json file is generated containing the filename, heading angle, and a confidence value.

The implementation is self-contained and does not affect existing modules. It is meant as an optional utility for generating training data for gate detection experiments.