Run manual QA tests

[karencfv]

Run several drift scenarios to make sure the provider runs as expected. The instance resource is really complex. What I want is make sure we don't have random dangling resources that aren't represented in the state during a botched run, and the users can always sync with the real state of things.

• Instance disk attachment deletion/modification from sources outside of the scope of TF
• Instance nic attachment deletion/modification from sources outside of the scope of TF
• Delete things (VPCs, subnets) randomly during creation/update/deletion
• Random other stuff?

[karencfv]

Blocked until dogfood is back up