Increase Workhorse CI long polling interval

CI runners that poll GitLab.com for new jobs are intercepted by long polling application logic in Workhorse. Most long polling request hit the long polling timeout which we have configured at 50s (the -apiCiLongPollingDuration flag).

Because each runner hits this endpoint in a loop, and because requests take 50s, each runner presents a workload of 1/50 = 0.2 requests per second.

If we were to increase that timeout, we reduce the runner request rate and thus reduce saturation of server resources. Most notably Redis.