This PR makes _RenderSlider not be a semantics container. This is so that the FocusableActionDetector in the Slider widget will get to aggregate the semantics information, since otherwise Talkback won't focus the slider because it thinks that the focus node doesn't have anything to say (which it doesn't but the _RenderSlider child does). If the _RenderSlider is a semantics container, then it keeps its speakable information to itself, but it isn't the Focus widget, so when the keyboard focus goes to the focus node, the accessibility focus doesn't move.

Since the _RenderSlider is always wrapped by the Slider widget, there's nothing lost in making it not be a container.