fix(handler/ec2-fleet): set resource_id on instant fleet machine stubs

When _acquire_hosts_internal builds the initial instance list for an
instant fleet, each stub only carried instance_id. Without resource_id,
group_by_resource (used by the return path) silently skips every stub,
so the machines are never fed to terminate_instances and the AWS
instances stay running until they expire or are cleaned up out-of-band.

Stamp resource_id=fleet_id on every stub at creation time. fleet_id is
already in scope from the line above where the fleet was created.

Author: fgogolli

src/orb/providers/aws/infrastructure/handlers/ec2_fleet/handler.py

@@ -155,7 +155,9 @@ def _acquire_hosts_internal(
                 # lazily by the check_hosts_status / _check_single_fleet_status polling path.
                 instance_ids = fleet_result.get("instance_ids", [])
                 if instance_ids:
-                    instances = [{"instance_id": iid} for iid in instance_ids]
+                    instances = [
+                        {"instance_id": iid, "resource_id": fleet_id} for iid in instance_ids
+                    ]
                     self._logger.info(
                         "EC2Fleet instant fleet created with %d instance(s): %s",
                         len(instance_ids),