Runner Managers should have Continuous Profiling Enabled

Suggested during production#4209 (closed), so marking as a corrective action

We run Stackdriver Continuous Profiling on our Go services.

For example, Workhorse Continuous Profiling is here: https://console.cloud.google.com/profiler;timespan=4h/workhorse-web;type=CPU/cpu?project=gitlab-production

We should configure and enable this for the runner manager fleet.

cc @tmaczukin @igorwwwwwwwwwwwwwwwwwwww @rehab @marin @rnienaber @cmiskell