[cloud deploy] Enable ML autoscaling (#188653)

Enables ML autoscaling on Cloud deployments, in support of ELSER models
requiring 4gb of memory.

.buildkite/scripts/steps/cloud/deploy.json

@@ -29,7 +29,6 @@
           "dedicated_masters_threshold": 6
         },
         "plan": {
-          "autoscaling_enabled": false,
           "cluster_topology": [
             {
               "zone_count": 1,
@@ -133,17 +132,22 @@
               }
             },
             {
-              "zone_count": 1,
+              "autoscaling_max": {
+                "resource": "memory",
+                "value": 8192
+              },
+              "autoscaling_min": {
+                "resource": "memory",
+                "value": 0
+              },
+              "autoscaling_tier_override": true,
+              "id": "ml",
               "instance_configuration_id": "gcp.ml.1",
               "node_roles": [
                 "ml",
                 "remote_cluster_client"
               ],
-              "id": "ml",
-              "size": {
-                "value": 1024,
-                "resource": "memory"
-              }
+              "zone_count": 1
             }
           ],
           "elasticsearch": {
@@ -208,7 +212,9 @@
       }
     ]
   },
-  "settings": {},
+  "settings": {
+    "autoscaling_enabled": false
+  },
   "name": null,
   "metadata": {
     "system_owned": false