Latency emulation

[mcantrell]

Typically there will be some overhead due to network and the work done by the remote service. In order to come closer to emulating a real remote system, the API should be able to emulate the latency involved with making the request.

Provide a mechanism to wait a given amount of time before sending a response to each thread.