Add Notification of Shared Resource Usage

This blueprint aims to address the problem of compute nodes incorrectly reporting resource amounts when they share the resource, such as when nodes use Ceph or NFS shared storage for their ephemeral disks. When an instance is created or destroyed on one of these shared-storage nodes, the other nodes are not aware of the change, and continue to report the old, stale values. This proposes to add notifications by the scheduler to the hosts that share resources whenever that amount is changed so that the compute nodes can update their reported resources accordingly.