Google Cloud regional workers and improvements

[dghubble]

• Change Google Cloud module to require the region variable (CHANGE)
• Worker managed instance group creates workers across zones in a region.
  • Example: region: us-central1 means a worker may go in us-central1-c, us-central1-f, us-central1-a, etc.
  • Tolerate Google Cloud zone failures or capacity problems
  • Tradeoff: Changing user-data will no longer automatically roll workers. You can delete old instances and they'll be replaced by new instances. Worthwhile tradeoff since mutating user-data should only be needed in special cases. See Add new resource region_instance_group_manager hashicorp/terraform-provider-google#394
• Improvement/Note for Preemptible VMs:
  • If worker preemption is enabled, replacement instances can now be drawn from any zone in the region, which should avoid scheduling issues that were possible before if a single zone aggressively preempted instances (presumably due to Google Cloud capacity).

Cleanups:

• Fix typo in internal template variable name: ssh_authorized_keys should be ssh_authorized_key to match the user facing variable which only allows a single SSH authorized key
• Remove controller_preemptible option: Controller preemption is not safe or covered in documentation. Delete the option, the variable is a holdover from old experiments (BREAKING)
• Require google provider plugin 1.1 or higher which includes fix: terraform plan always finds a disk device_name diff hashicorp/terraform-provider-google#574. Allows us to remove the static persistent disk name workaround (CHANGE)