Ability to pop out video tiles ("picture-in-picture")

One request that we've been getting recently is a way to activate the browser's [picture-in-picture functionality](https://developer.mozilla.org/en-US/docs/Web/API/Picture-in-Picture_API) from within Element Call.

Yet to be discovered:
- What are the use cases for this?
- Would we rather provide a feature that lets you pop out a specific tile only, or one that creates a window always highlighting the active speaker?
- On the implementation side: how does this interact with LiveKit automatically unsubscribing from video streams that are off screen?
- Is this worth implementing given that Firefox doesn't yet support the relevant APIs?
- Design (potential for integrating it with the drag-and-drop interaction, such that you could literally drag a tile and drop it outside the window to pop it out, much like a browser tab? or do we need something less surprising like a button)