With some recent refactoring I broke animation-auto-play. Previously it was doing some juggling with responsibilities split across the Viewer and the ViewerElement. With this change, I moved all the responsibility to the Viewer, dependent on load options. This makes it much simpler to ensure the logic is correct. An alternative would be to make the ViewerElement entirely responsible, and have it sequentially load the model, then set the selectedAnimation, then play the animation. The only reason I don't want to do it this way is because changing the selectedAnimation animates the camera pose (based on computed bounds), and when the model is first being loaded, we don't want this animation. So it's either add load options for the Viewer, or expose the selectAnimation function so the ViewerElement can select an animation without animating the camera pose. But since other view framework layers would want the same behavior, I think it's better to do it in the Viewer.