If you are interested, there is also a spec under development at the moment to make it more explicit - https://wicg.github.io/picture-in-picture/