If `setTimeout` is overridden in an async test, ospec's timers can break, causing ospec to not actually report anything.

[dead-claudia]

Steps:

1. In either a o.beforeEach callback or synchronously in a promise-returning test body, override setTimeout to capture its body.
2. In the test body, asynchronously invoke the callback at some point.
3. Restore the callback either at the end of a test or in an o.afterEach callback.

Unfortunately, I don't have a handy test case for this at the moment, but it should be possible to piece one together pretty easily.

Had to work around this by using timers directly.