You're describing a feature that we've discussed several times before (the ability to output to multiple devices simultaneously). There are some serious technical challenges getting this to work in all scenarios (for instance if you're outputting to an AV receiver and headphones, the sound may be several milliseconds out of sync which sounds crappy) and as such it hasn't made the bar for features (there have always been higher priority features to implement).
In Win7 when you plug your headphones in, audio will be automatically redirected (assuming that the app hasn't explicitly said that it wants to render to your desktop speakers), and when you unplug them it will be automatically redirected to the desktop speakers. We've done a lot to make sure that everything works smoothly but we've not yet implemented all the features we want.
Part of the reason that I use weasel words like "probably" and "maybe" is because these features DON'T work in all circumstances. For instance the stream switching behavior intentionally only works if an app uses the default output device - that's because we're not going to override the choice of the application - if the application said that it wants to render to the speakers, we're not going to override the app's choice.