manager: race condition with watchdog while starting workers
How to reproduce:
- disable worker count limit
- set worker count to 1000
- run the manager
- watch the world burn (like really, these steps will trash your system)
- manager crashes (on my machine after starting 183 instances of kresd)
I am guessing, that the same behavior could be reproduced by hammering manager with worker count change requests. But I haven't tested that.