Remove reason from Events, add session

[toji]

The idea of stating a reason with VR events was initially brought up to allow things like the activate event to differentiate between activating because a proximity sensor was set off or a page was navigated or a phone was placed in a harness, etc. Upon reflection I question the utility of this, though. We're already seeing that in the 2.0 spec we need separate events for navigate and activate, and it's not clear that developers will benefit from being able to differentiate between other reasons for activation. For most other events (like connected or blur) the reason ended up simply being a repetition of the event name itself, which isn't useful.

For this reason I propose that in 2.0 we drop the reason attribute on the VR events. Additionally, presuming that we go with the VRSession concept in #179, it DOES seem useful to communicate the session that an event is associated with (if there is one). This means the IDL would end up looking like this:

interface VRDisplayEvent : Event {
  readonly attribute VRDisplay display;
  readonly attribute VRSession? session;
};

Alternatively, we could have separate event types for a display and session, like so:

interface VRDisplayEvent : Event {
  readonly attribute VRDisplay display;
};

interface VRSessionEvent : Event {
  readonly attribute VRSession session;
};

But this feels overly complicated to me without a clear benefit.

[NellWaliczek]

Thanks for cleaning this up! I've got a slight preference for the latter option; it seems a bit cleaner. The question is for events off of navigator.vr, do we need a VREvent type? My hunch is no, since the connected/disconnected events will still need a VRDisplay as a parameter.
Thoughts?

[toji]

I don't think there's any vr-centric events that don't have a related display or session, so I don't think that's needed.

Related, though I didn't want to tackle it in this particular issue: I'm going to propose renaming it to VREvent anyway to prevent some naming conflicts between 1.1/2.0 [Edit: Filed #192 to address this.]

[RafaelCintron]

I am also in favor of having separate event types. If we have the session point back at the display, the web developer can go from one to the other.

[toji]

Talked with some of the platform leads on Google's end and it seems we have a slight preference for two event types in this situation as well, so I'm going to update the explainer to reflect that.

[toji]

Closing since it seems like we've come to a general consensus on this, which is now reflected in the explainer.

If anyone has further issues with the shape of WebVR's event interfaces please open a new issue.