refactor: remove spec sleeps

[caspiano]

There were a bunch of sleeps littered through specs.
I think we should try to avoid these entirely.
if they are necessary, we should implement some kind of asynchronous abstraction or make requests synchronous.

[kimburgess]

Responses from exec's can be awaiting by calling #get on their result, which blocks until the Future resolves. One that I've defined run into (and had to insert a sleep to accomodate) is the initial init / settings load.

This change is definitely the correct thing to do, but will need to be synchronised with improvements to spec-runner.