'pauseTest' doesn't reset the test timeout

[azhiv]

In #291 the pauseTest method was updated in order to ignore the test timeout. However, this doesn't work as expected - putting await this.pauseTest(); somewhere in the test body doesn't actually prevent qunit to fail the test due to the timeout being hit. I found two workarounds:

1. put assert.timeout(-1); in the beginning of the test
2. run the test in devmode

The thing is none of this options is stated in the corresponding comment in addon-test-support/ember-qunit/index.js. This is a very convenient feature, so it would be really great to either update the comment or fix the behaviour to match the expected one.

[rwjblue]

Thank you for reporting!!! I took some time to dig into this issue the other day, and determine that the issue here is that when the test is an async function we have already returned a promise and qunit has already read the timeout value before you have a chance (in the test body) to set assert.timeout(-1).

Specifically, this is the order of events today:

• QUnit invokes the test body (which is an async function) here
• A promise is returned (because that is how async functions work) so Test.prototype.resolvePromise falls into the promise handling code here
• Before chaining off the promise (to pass/fail the test based on success or failure) QUnit uses the current timeout value to set up a race against the promise resolution here
• Only after all of that is the actual test body going to be invoked
• Sometime later, the user calls await pauseTest() which calls this.pauseTest() here which ultimately invokes assert.timeout(-1) here
• assert.timeout sets testInstance.timeout to the new value here, but never cancels/interacts with the timer created above by returning a promise from the test body

---

Proposed solution:

We need to update QUnit's implementation of assert.timeout to reset any current timeouts by canceling and setting up a new timer (if the timer is > 0).

[rwjblue]

Writing all that up made me realize there is a temporary work around we can do until we get the fixes in QUnit: #497.

[step2yeung]

Added a PR on QUnit for this! cc @rwjblue @scalvert

[rwjblue]

The fix from @step2yeung landed in qunit@2.9.3.