[css-animations] reverting an earlier change about animation-timing-function

I was looking at an issue in css-animations with Cameron, and
spotted something in the animations spec that seemed wrong to me.
It went back to this edit:
  https://hg.csswg.org/drafts/rev/685f66d955ec
which introduced text saying that an 'animation-timing-function'
within a keyframe rule would apply until the next keyframe rule that
had another animation-timing-function declaration.  Previously I'd
have interpreted the spec as implying that it was until the next
keyframe for the property being animated, whether or not that
keyframe has an animation-timing-function declared.

I'm not sure why this edit was made.  However, based on this test:
  http://dbaron.org/css/test/2015/animation-timing-function
it disagrees with all implementations that I've tested (Firefox,
Chrome, IE11 [1]).  Given that the edit was intended as a
clarification, I plan to change the spec to re-clarify things in the
other direction, i.e., to match the way all implementors that I'm
aware of had interpreted the old wording.

-David

[1] Note that IE11 has a bug going in the reverse direction in
    animation (C), but it agrees with the others for the forward
    direction.  I wasn't able to test Edge since the VM with Edge
    seems to be busted for me right now.

-- 
𝄞   L. David Baron                         http://dbaron.org/   𝄂
𝄢   Mozilla                          https://www.mozilla.org/   𝄂
             Before I built a wall I'd ask to know
             What I was walling in or walling out,
             And to whom I was like to give offense.
               - Robert Frost, Mending Wall (1914)