ViTPose++: Vision Transformer for Generic Body Pose Estimation