Fixed issue where an unexisting instance could not be destroyed.

[jlesage]

I'm using the autoscaler with DigitalOcean. The autoscaler went into a state where the server count was not reflecting the reality. For example, when no Droplet existed on my account, the autoscaler was reporting something different:

9:12AM DBG check capacity id=Bjc93TuZjLNlUqoc max-pool=4 min-pool=0 pending-builds=0 running-builds=0 server-buffer=0 server-capacity=4 server-count=2

Attempting to remove these servers with the cli via server destroy command results with the following error on the autoscaler:

11:03AM DBG destroying server server=agent-p8clg4bK
11:03AM WRN cannot stop the agent error="Cannot connect to the Docker daemon at https://134.209.214.159:2376. Is the docker daemon running?" server=agent-p8clg4bK
11:03AM WRN droplet does not exist error="GET https://api.digitalocean.com/v2/droplets/141608491: 404 The resource you were accessing could not be found." image=docker-18-04 name=agent-p8clg4bK region=nyc1 size=s-4vcpu-8gb
11:03AM ERR failed to destroy server server=agent-p8clg4bK

This PR fixes the removal of these servers by looking at the error, similarily to what is done in engine/reaper.go.

[bradrydzewski]

thanks!