Update JaniceMediaIterator to support multiple images

[mlngan]

Operationally, there's a use case for generating a template from a pile of images of the same person, obtained from different sources. Right now, the documentation for JaniceMediaIterator supports a single image or a video. Can we update JaniceMediaIterator to support multiple images (of different sizes) of the same person? That way, it'll be a single call to janice_enroll_from_media() rather than multiple calls to janice_io_create_media_iterator_from_image(), janice_detect(), then janice_enroll_from_detections().

[JordanCheney]

@mlngan I think the new sparse media iterator can do what you need. The constructor is here- https://github.com/Noblis/janice/blob/master/implementations/opencv_io/janice_io_opencv_sparse.cpp#L172

The documentation should probably be updated to say multiple images or multiple frames from a single video. In either case I think the interpretation is the same- i.e. find the person or persons who appear (most likely multiple times) across this collection of media without tracking.