Feature detecting outerWidth setter API expensive

I suppose we can infer based on the function length, but I'd really prefer to keep the real feature detect. Ideally we'll get this split out into a file for supporting old jQuery Core and then you can just exclude that in your build.

I'll leave this open until we can discuss it with the whole team.

Can you provide the test page you're using that's showing ~25ms reflow time? My test page is admittedly simple, but I'm only getting ~1ms time spent on the detection, ~2ms total time spent on detection + implementation. I'm even seeing similar performance in IE8.

Due to having jQuery UI in the head, and early on in our scripts, this triggers one of our first recalculate style / layout / rendering.

The cost of that reflow should be proportional to the size and complexity of the page. The 1ms / 2ms numbers you quote for a test page make sense to me, it's only on the large DOM we're dealing with that would cause the 20ms+ reflow time.

It's a little tricky to distill a meaningful test page that shows it - but it can be seen clear as day on one of our production systems. It's about 20ms on this page as seen in the chrome dev tools timeline: https://confluence.atlassian.com/404.action and here's a screenshot of the dev tools https://s3.amazonaws.com/uploads.hipchat.com/10804/51242/J7tb07lp7m6H7yj/Screen%20Shot%202015-02-27%20at%2010.48.43%20am.png

Core: Drop feature detection for version check to support jQuery 1.7

Fixes #11197

Closes gh-1458

Changeset: c077e0a347826c7acca4ca2f775ca59e3b5f8d07