Added support for angular and linear velocity

[cabanier]

Closes #619

Also silences a couple of bikeshed warnings.

---

Preview | Diff

[cabanier]

@Manishearth why can I not add reviewers?

[toji]

LGTM after one correction.

[toji]

GitHub had a hangover this morning, which is why you couldn't add reviewers. I think it's generally better now.

[cabanier]

GitHub had a hangover this morning, which is why you couldn't add reviewers. I think it's generally better now.

I still can't add reviewers.

[cabanier]

I was looking into a possible implementation and there are some spaces where velocity will never be reported.
For instance XRJointSpaces and XRTargetRaySpace with a XRTargetRayMode different than POINTING.

Should we add those exceptions to the spec, or leave it up to the UA?

[toji]

I think it's best to leave that up to the UA. For example, I can see that OpenXR has an extension for hand joint velocity and I suspect that the tracking method (ie: optical vs sensors) will make a big difference in what velocities are available. I wouldn't want to prematurely prevent future advances in tracking.

If we're concerned about developers needing to know what spaces can and can't report velocities, it would probably be best to report that as a property of the space (hasLinearVelocity/hasAngularVelocity booleans, maybe?) but I don't know that we can do that consistently and reliably on all platforms, and I believe even spaces that CAN report velocities may have moments where it lapses, such after the first frame or two of tracking (depending on the tracking method). As a result, developers will need to be able to handle null velocities no matter what.

We should probably just leave it as "If the user agent can't populate this, it's allowed to return null."

[toji]

Thanks for fixing the angular velocity definition, by the way!

[cabanier]

@toji thanks for approving! Can you merge it for me? I'm not able to.

[toji]

If you don't mind I'd like to make sure that @Manishearth has had a chance to look before I do? That should be Tues at the latest, since we have our editors sync then. Normally I wouldn't be too concerned but given this is a change to a largely "shipped" spec I feel a smidge of extra caution is warranted.

I doubt there's going to be any concerns, though.

[cabanier]

I see. That sounds good! There's no hurry to get it in.

[Manishearth]

Seems good now.

One note: angular velocity can be obtained from the other attributes; and we tend towards not having APIs that expose information that can be calculated by the author, but I think it's fine in this case.