Remove deprecated asyncTest usage in favor of assert.async for QUnit v1.16.0+

[JamesMGreene]

The forthcoming QUnit v1.16.0 will deprecate QUnit.asyncTest in favor of a new assert.async() approach (see qunitjs/qunit#653). This plugin should be updated to utilize this style of test as well.

More importantly, this plugin must operate in a very non-standard way that could make it inviable to use the new assert.async() approach as a replacement for its current use of QUnit.stop/QUnit.start. We should confirm that soon-ish as well in case we need to take an alternative approach... most likely we will just need to wrap the done callback in an indirect but globally available closure so it can be invoked when the child frame's test run finishes (i.e. as a replacement to qunit-composite.js#L86-L87).

[platinumazure]

I might be able to explore this in the next couple of weeks, if you're willing to accept a PR.

[platinumazure]

Oh, question. Are you aiming for backward compatible support of pre-1.16, or are you okay with dropping support for that with this change (and possibly doing a major version bump)?

If you did that, you could probably also not bother with the complicated fix to QUnit.push, either.

[platinumazure]

Pinging @JamesMGreene. Can you please answer this question when you get a chance?

Oh, question. Are you aiming for backward compatible support of pre-1.16, or are you okay with dropping support for that with this change (and possibly doing a major version bump)?

Thanks!

[JamesMGreene]

Sorry for the [now usual] delay, Kevin. In all honesty, with the days of QUnit 1.x coming to an end and QUnit 2.0.0 forthcoming, it's probably best to just hold off on making any change like this until QUnit 2.x hits the scene. At that point, we can do a major version bump of this module as well and align it with QUnit 2.x.

[platinumazure]

@JamesMGreene Works for me.

[jzaefferer]

You could test against 2.0.0-rc1: https://github.com/jquery/qunit/releases/tag/2.0.0-rc1 - not expecting any relevant changes till the final release.

[platinumazure]

@JamesMGreene Now that 2.0.0 is out, should we look into this again? I think the simplest solution is a major version bump of this module and a full breaking change to just go into full QUnit 2.0 compatibility mode.

[JamesMGreene]

@platinumazure: Yessir, that would be best.

[platinumazure]

Thanks @JamesMGreene. I'll see if I can come up with a PR in the next couple of weeks, unless you beat me to it.