Analyze autoprefixer targets

Our autoprefixer browser targets currently sits like this:

``` js
[
'ie >= 10',
'ie_mob >= 10',
'ff >= 30',
'chrome >= 34',
'safari >= 7',
'opera >= 23',
'ios >= 7',
'android >= 4.4',
'bb >= 10'
];
```

We should review these numbers based on our target support and some existing usage analytics. For example, Chrome 34 is quite dated (up to like 49 in dev channel.) Seems safe focusing only Chrome-wise to update this to around 45 (considering current stable is 47, give a little headroom.)

Our Android support is also targeting 4.4+. Should we look into handling android issues that go back a little further (as we already have with the menu icon being problematic in older webviews)? If so, that should be pushed back or we need to make it clear which Android webview versions we support internally.

Firefox is currently at version 43. 38 is the current ESR version with 45 being the next up. At the time the update of browsers happens, we should use the latest ESR version for a support target. This way we are targeting the slowest upgrade path of FF while still staying fresh enough to hopefully not bloat the output with uneeded prefixes.