Support and tests for vrdisplaypointerrestricted events to the a-scene…

[olga-microsoft]

…element with the possibility to disable the behavior (on by default)

Description:
Support mouse input in WebVR by handling vrdisplaypointerrestricted and vrdisplaypointerunrestricted events in MS Edge. It is by design that mozPointerLock* and webkitPointerLock* implementations are not included.

Changes proposed:

• In response to vrdisplaypointerrestricted perform requestPointerLock on the canvas element that is used for WebVR rendering
• In response to vrdisplaypointerunrestricted perform document.exitPointerLock if the locked element is the canvas that is used for WebVR rendering
• Give the developer an option to turn off this behavior by calling disableCanvasPointerLock on the scene object.

[dmarcos]

What are the scenarios where someone would want to enable / disable this? I don't fully understand Edge WebVR behavior. I'm poking on this with partial information.

[olga-microsoft]

Just wanted to be flexible. I can't think of a scenario that this would be useful. I will remove this if it's okay with you.

[dmarcos]

As a maintainer I'm becoming more and more flag averse over time 😄 since new flags combine with previous ones that can interact in multiple weird ways that are difficult to debug and maintain over time. As a conservative measure I would get rid of the flag now and reintroduce it later if we found a legitimate use case for it. I cannot think of any at the moment.

[olga-microsoft]

Sure, this makes sense, removing!

[dmarcos]

Thanks!