Encode ``EditedMediaItem``s as a motion photo

Use case description

Currently, the media3 Transformer APIs allow you to either use the photo or video portion of the Motion Photo as an EditedMediaItem.

This is the opposite of what I need to do. My use case is combining a sequence of images (EditedMediaItems) and encoding a Motion Photo. The image sequence would be encoded as the video portion and one of the images would be set as the photo portion.

Proposed solution

Provide and API to encode Motion Photos from EditedMediaItems.

A naive take on what the API could look like:

Transformer.Builder(context)
    .setMotionPhotoMimeType(MimeTypes.IMAGE_JPEG)
    .setMotionPhotoPrimaryPhoto(EditedMediaItem(...))
    .setMotionPhotoMediaItems(ImmutableList.of(
        EditedMediaItem(...),
        EditedMediaItem(...),
        ...
    )).build()

Alternatives considered