Requested feature

Currently, if image captioning is turned on by pipeline_options.do_picture_description = True and the extracted markdown is saved with the document.save_as_markdown(output_md_name, artifacts_dir=assets_dir, image_mode="referenced") method, the captions are inserted above the image reference.
It would be great if one could combine them so that for each image, the caption would be inserted into the reference, like ![image_caption](path_to_image)