Feature request: Render captions inside the image to support Media Viewer caption display

Currently, `<caption>` does not render the caption inside an image, only as text below the image or other media. MultimediaViewer does not support this as it expects a caption inside the image *a la* `[[File:filename|caption]]`. It would be beneficial to have the contents of `<caption>` (safely) embedded in the media itself if possible.