Add restart and resize actions for cloud nodes

Found by Task 12152 (OOM loop on jaah-node-pool-c-host). A Multi-Node Host stuck in a memory-starved state cannot be recovered from the console: there is no action to reboot the host's EC2 instance or change its instance type, so the only routes are the AWS CLI/console, which the hub's "operate Jaah only through the UI" rule forbids. - Add a host-level "Restart host" action on a cloud node (its machine, not just its daemon), confirmed, audited, refused while the CP cannot tell whether slots hold live work unless the operator confirms. - Add "Change instance type" for a CP-launched host (stop → modify → start), limited to the allowed instance-type list (`cloud_node_launch.require_allowed_instance_type`). Out of scope: node groups' scaling rules; any change to the vnode memory ceiling (packaging/jaahvnodes.slice). Done means: - An operator can reboot a cloud host and change its instance type from the node's page; both are audited and covered by API tests.

Added Restart host for cloud nodes. It is confirmed and audited, and refused while live work or silent slots can't be ruled out unless forced. Allowed on group members, and for the operator on Hosted machines. Change type already existed and now has an audit sentence. Quick gate and changed-area tests green; PR #7577 queued in the merge lane. Memory: none applied.