* object callbacks

per previous request

added after call
added arguments to both calls

* faster function evaluation, missing comma

getElapsedTime() in TREE.Clock returns whole running time and not elapsed time from previous start. (#9692)

* getElapsedTime() returns whole running time

getElapsedTime() function returns incorrect value after stop() and start(), It returns whole working time from first run.
I changes adde this.elapsedTime = 0 in stop() function, so after next start() we can get correct elapsed time from last start()

Perhaps it's incorrect and suggested changes need new function, so Clock.js will have two functions:
- first one will return whole running time (as it is now)
- second will return elapsed time from last start

* Update Clock.js

* PolyhedronBufferGeometry: Initial commit

* PolyhedronBufferGeometry: Update docs

Reading clientHeight of canvas during every rAF causes layout thrashing and results in frame drops. Instead re-use the _height local variable available in current scope.

Corrected the spelling of argument in docs/manual/introduction/Creating-a-scene.html

- Move computation from fragment shader to vertex shader wherever possible
- Eliminate some dead shader code and comments
- Fix spelling of "Rayleigh"

issue #9739

The spec has been updated so that `requestPresent` can accept empty arrays for `leftBounds` and `rightBounds`. Though Chromium returns `null` for these properties by default, Firefox returns empty arrays, and there is a proposal to make those non-nullable. This patch should fix rendering for Firefox, which is currently broken.

For reference:
https://github.com/w3c/webvr/issues/95
https://github.com/w3c/webvr/issues/89

The spec has been updated so that `requestPresent` can accept empty arrays for `leftBounds` and `rightBounds`. Though Chromium returns `null` for these properties by default, Firefox returns empty arrays, and there is a proposal to make those non-nullable. This patch should fix rendering for Firefox, which is currently broken.

For reference:
https://github.com/w3c/webvr/issues/95
https://github.com/w3c/webvr/issues/89

* BufferSubdivisionModifier: cleanup

* BufferSubdivisionModifier: More cleanup

They didn't produce correct eye translations in all cases (specifically
standing scenes.) This reverted version may not be 100% accurate on
future hardware (if the eyes have a rotation component, for example) but
it's correct on all current WebVR implementations. That's better than
blatantly wrong in some cases.

Use of the WebVR-provided projection matricies has been left in, since
that was working correctly.

Fix Safari glTF rendering issue re: WebGLRenderingContext constant values are not exposed (see: https://github.com/mrdoob/three.js/issues/8973) (#9721)

They didn't produce correct eye translations in all cases (specifically
standing scenes.) This reverted version may not be 100% accurate on
future hardware (if the eyes have a rotation component, for example) but
it's correct on all current WebVR implementations. That's better than
blatantly wrong in some cases.

Use of the WebVR-provided projection matricies has been left in, since
that was working correctly.