Comment 7 for bug 1250168

It looks like increasing the timeout not only confirms Salvatore's assessment, but it is inline with how the fact that the gate started breaking after this change merged (in fact we get a failure in this very change). Obvious short-term solution would be to revert change I7e7c767400ca448cb86d27b60a1229a2afa69726, or even better customize the BUILD_TIMEOUT to 400 (the old value) for the large-ops job until we can figure out a more permanent solution.