"Run" request to server always takes maximum allowed time (macOS)

[ots22]

When running on macOS, a run submitted via the server always takes the length of time specified by the nanosleep in sheep-server.cpp, even if running the circuit takes much less time. (To reproduce, submit any simple circuit calculation using the 'clear' context via the web interface, which should return very quickly; instead this returns successfully after 10 seconds, when the timeout is reached).

The expected behaviour is for the request to return as soon as the run completes. The process spawned to handle the request returns SIGCHLD to the parent server process, which should interrupt the nanosleep but does not seem to on macOS. It was verified that SIGCHLD is received by the parent, however 'nanosleep' returns 0, indicating it was not interrupted.