Change I0839470c4bcfb16590a0d87b306d683b059bf8a9 fixed
root disk usage tracking for volume-backed instances
when performing an instance_claim which happens during
the initial server create and unshelve. However, root
disk reporting is still wrong for volume-backed instances
during move claims (resize and evacuate) because a move
claim calls _update_usage_from_migration which passes a
Flavor object to ResourceTracker._get_usage_dict() and
that method didn't have "is_bfv" logic in that scenario.
This fixes the bug by always passing the instance object
to the _get_usage_dict() method so we can determine if
it's volume-backed and if so report the root_gb usage as 0.
The related functional regression test is updated
appropriately to show the bug is fixed for volume-backed
instances.
Change-Id: Ia19264ae7c88bb03ed3118795d4011ceb62ef92c
Closes-Bug: #1796737
(cherry picked from commit a99722bb85)