Retry on flaky test failures

- indices/open.yaml: Add retry to _close prologue steps to handle
  transient 400 when closing a freshly created index before shards
  are fully initialized.
- terms_aggregate_unsigned_long.yaml: Remove size limit from format
  chapter so all buckets are returned (avoids version-dependent
  sum_other_doc_count) and drop key_as_string assertions that vary
  by precision handling across versions.
- ml/models.yaml: Restore retry on "Search model" chapter since model
  indexing has a delay after registration completes.

Observed on: OpenSearch 2.19.5 and 3.6.0

Signed-off-by: Sean Chittenden <sean.chittenden@crowdstrike.com>

Author: sean-

tests/default/_core/search/aggregations/terms_aggregate_unsigned_long.yaml

@@ -150,7 +150,6 @@ chapters:
           metric_ids:
             terms:
               field: metric_id
-              size: 2
               format: '0,000'
     response:
       status: 200
@@ -161,10 +160,8 @@ chapters:
             sum_other_doc_count: 0
             buckets:
               - key: 9223372036854775808
-                key_as_string: '9223372036854776000'
                 doc_count: 2
               - key: 12345678901234567890
-                key_as_string: '12345678901234567000'
                 doc_count: 2
 
   - synopsis: Aggregate by unsigned_long field with order by count and verify max uint64 value.

tests/default/indices/open.yaml

@@ -15,12 +15,21 @@ prologues:
   - path: /movies/_close
     method: POST
     status: [200]
+    retry:
+      count: 3
+      wait: 1000
   - path: /games/_close
     method: POST
     status: [200]
+    retry:
+      count: 3
+      wait: 1000
   - path: /films/_close
     method: POST
     status: [200]
+    retry:
+      count: 3
+      wait: 1000
 epilogues:
   - path: /movies,games,films
     method: DELETE

tests/plugins/ml/ml/models.yaml

@@ -48,6 +48,9 @@ chapters:
         hits:
           hits:
             - _score: 1
+    retry:
+      count: 3
+      wait: 5000
   - synopsis: Get model.
     path: /_plugins/_ml/models/{model_id}
     method: GET