I just hope Web Intents makes it into the actual release - and in a way that is usable by normal users. Seems to me this should be a priority, instead of putting all that effort into yet another single sign-on fantasy.
If this happens as part of a large integrated experience as you promised that's fine, but I fear that you are taking on too much, going for a complex and over-engineered package that may never be ready or actually usable, instead of concentrating on this one single important thing which really needs to get done.
I used to share your skepticism about single sign-on. I'm actually quite bullish about BrowserID because it uses concepts that both users and websites are very familiar with (email, 'validated email addresses'), and works it in a flow that feels very much like facebook connect. It's really neat rather than overengineered IMO (and I can say that because I didn't do any of the engineering ;-).
Similarly, the apps project is a lot about recognizing what mobile apps did well, and filling in the gaps in the web (stores, receipts, APIs) that will make it possible for people to translate their mobile appdev skills into the broader web.
If this happens as part of a large integrated experience as you promised that's fine, but I fear that you are taking on too much, going for a complex and over-engineered package that may never be ready or actually usable, instead of concentrating on this one single important thing which really needs to get done.