WAON: Large-Scale and High-Quality Japanese Image-Text Pair Dataset for Vision-Language Models