Bugzilla

Per Hiro's comment in the patch, we should check this testcase as well.

I saw something like this in Chrome/Safari:
```
transitionrun(opacity) on 770.9
transitionrun(transform) on 770.9
transitioncancel(transform) on 1270.8
transitioncancel(opacity) on 1270.8
transitionrun(transform) on 1270.8
transitionrun(opacity) on 1270.8
```

but in Gecko:
```
transitionrun(opacity) on 756.56
transitionrun(transform) on 756.56
transitionrun(transform) on 1289.8
transitioncancel(transform) on 1289.8
transitionrun(opacity) on 1298.14
transitioncancel(opacity) on 1298.14
```

In this case, we did cancel transform transition and create a new one to replace it first, but Chrome/Safari fired the transitioncancel of transform and opacity together. This is probably implementation-dependent, and maybe we just need to make sure `transitioncancel({transform|opacity})` is earlier than `transitionrun({transform|opacity})`, correspondingly.

Per Hiro's comment in the patch, we should check this testcase as well.

I saw something like this in Chrome/Safari:
```
transitionrun(opacity) on 770.9
transitionrun(transform) on 770.9
transitioncancel(transform) on 1270.8
transitioncancel(opacity) on 1270.8
transitionrun(transform) on 1270.8
transitionrun(opacity) on 1270.8
```

but in Gecko:
```
transitionrun(opacity) on 756.56
transitionrun(transform) on 756.56
transitionrun(transform) on 1289.8
transitioncancel(transform) on 1289.8
transitionrun(opacity) on 1298.14
transitioncancel(opacity) on 1298.14
```

In this case, we did cancel transform transition and create a new one to replace it first, but Chrome/Safari fired the transitioncancel of transform and opacity together. This is probably implementation-dependent, and maybe we just need to make sure:
`transitioncancel({transform|opacity})` is earlier than `transitionrun({transform|opacity})`, correspondingly.