Refactored the information it contained into VRFrameData (offset is now
*PoseToEyeMatrix for more flexibility) and the renderWidth/Height moved
onto the VRDisplay as recommendedRenderWidth and
recommendedRenderHeight. They now describe the ideal canvas size,
containing both eyes, rather than the ideal render target size for a
single eye. Field of view was already deprecated and replaced with the
projection matrix on VRFrameData.

Removed getPose, because if we’re gonna break backwards compatibility
let’s go all in and get rid of “deprecated” functions that never
officially launched.

Fixes #126