PR #831 broke stale host cleanup for hosts that are part of groups

[mignaulo]

We recently upgraded our Teletraan setup to the latest commit from master and ran into issues where hosts stopped getting cleaned up when they where terminated. We were able to revert to the commit just before #831 was merged to restore this functionality. I'm no expert here, here is what I found out by digging into the PR.

The previous version of the cleanup code was split into two parts:

• Cleaning up hosts that are individually assigned to environments
  Previously handled by a function named processIndividualHosts, this function was renamed to processAllHosts. (https://github.com/pinterest/teletraan/pull/831/files#diff-d325e9aa7284f0411829cb75934a0ddae12c7281fda3b09da7779c406e8bad3cR96)
  The query backing this function has not changed much. It involved joining hosts to hosts_and_envs - it now involves hosts_and_agents joined to hosts_and_envs.

• Cleaning up hosts that are part of groups, those groups are assigned to environments
  This was previously handled by a loop calling the processEachGroup function on each group (https://github.com/pinterest/teletraan/pull/831/files#diff-d325e9aa7284f0411829cb75934a0ddae12c7281fda3b09da7779c406e8bad3cL124)
  This functionality was removed by PR 831. I guess it was assumed that the processAllHosts function would handle these hosts, however it only handles hosts that are directly attached to environments - not hosts that are part of groups.

Would it be possible to reintroduce the "cleanup stale hosts that are part of groups" functionality?
(Roughly these lines https://github.com/pinterest/teletraan/pull/831/files#diff-d325e9aa7284f0411829cb75934a0ddae12c7281fda3b09da7779c406e8bad3cL124-L173)

Thanks!

[mignaulo]

@ankilosaurus apologies for pinging you directly, but you were the author of PR831, which introduced the problem. Would it be possible to reintroduce the functionality that was mistakenly taken out?