Implement Animation.id

Clean up easings before parsing

Property names should be specified using camelCase, for examplebackgroundColor instead of background-color.

Use of hyphens in property names is deprecated

Don't store Javascript functions on AnimationEffectTiming.easing or keyframe.easing

Spec:https://drafts.csswg.org/css-values-3/#anglesIssue reported in bug#176 and bug #355

Matrix decomposition: support angle units grad and turn

Remove double call to normalizeTimingInput()

Throw TypeErrors on invalid inputs to AnimationEffectTiming

The polyfill evaluation of cubic Bezier timing functionsgives slightly different results than the native Blinkimplementation.In my testing, this change reduced the absolute error by 63%while only increasing the number of cubic function evaluationsby 31%Thiscloses#139

Increase Bezier timing function accuracy

The object-form-keyframes source file uses a call to Element.animateto decide whether to load the polyfill for the object-form syntax.If the native browser code supports Element.animate without fullsupport for object-form syntax, this test can throw a TypeError (see#72). Inthis case, the polyfill should proceed to load the polyfill. Thispatch implements this behaviour with a try-catch guard around theElement.animate loading check.

In response to review comments, this patch rewrites the test animationin the object-form-keyframes load check, to be more definite that itwill leave the element opacity untouched regardless of the outcome.Since the logic is now not obvious, a lengthy explanatory comment isalso added.

Add try-catch guard to object-form loading check

Fix release instructions

The property _isFinished is readonly. cancel() should not attempt towrite to the property.resolves#160

We make clear that old Chrome versions are not supported,and mention the option of use a requestAnimationFrame polyfill.

`npm install`was giving the warning`npm WARN deprecated minimatch@0.2.14: Please update to minimatch 3.0.2or higher to avoid a RegExp DoS issue`We avoid the warning by using a recent minimatch version.

Update minimatch dependency

Do not write to readonly _isFinished

Document that we require requestAnimationFrame

Currently, the semantics of [line 59 in `Gruntfile.js`](https://github.com/web-animations/web-animations-js/blob/ae52749433415fe979396bb82e7e5a0bdf9a115a/Gruntfile.js#L59) is - expose the library as a global called `true`. This causes two problems:1. The wrapping mechanism of uglify breaks in strict mode (more below)2. There's a leaking global variable called `true`The used version of uglify wraps the code in the following construct:```js"(function(" + parameters.join(",") + "){ '$ORIG'; })(" + args.join(",") + ")";```The IIFE passes the global `this` to a function and later tries to set the property name that we've assigned to `wrap`. This works fine in non-strict cases. The only side effect is the global `window.true` that we set to an empty object. In strict mode, however, global `this` equals to `undefined`. We try to set `undefined['true']` which cases a runtime error.We noticed the above scenario after enabling differential loading with `script[type="module"]`, which uses strict mode by default.It'll be fantastic if we can release the fix today, so we don't block the CLI release.Side note, the wrap function of uglify is fixed for strict mode [here](https://github.com/mishoo/UglifyJS2/pull/1811/files#diff-da84828c4bd7834c0040b0bbb51acdec). Maybe update of `grunt` and `grunt-contrib-uglify` is worth it.

Fix the wrapping IIFE of the prod bundle

…r/gui-and-headless-browsers/

Get Travis testing again