Prefetch all collection items globally before tests

Instead of making 3 GraphQL API calls per collection (~300 total for
100 collections), collect all repo and user references across all
collections upfront, deduplicate them, and batch the GraphQL queries.

This reduces API round-trips from ~300 to ~10-15 (depending on total
unique items), significantly improving test suite performance while
maintaining the same validation coverage.

Author: kenyonj

test/collections_test.rb

@@ -137,6 +137,8 @@
       end
 
       it "fails if a user, organization, or repository has been renamed or removed" do
+        prefetch_all_collection_items!
+
         errors = []
         repos_to_check = []
         users_to_check = []
@@ -151,10 +153,6 @@
           end
         end
 
-        cache_repos_exist_check!(repos_to_check)
-        cache_users_exist_check!(users_to_check)
-        cache_orgs_exist_check!(users_not_found_from(users_to_check))
-
         repos_to_check.each do |repo|
           repo_result = client.repository(repo)
           current_name_with_owner = repo_result&.full_name

test/collections_test_helper.rb

@@ -106,3 +106,43 @@ def annotate_collection_item_error(collection, string, error_message)
 def possible_image_file_names_for_collection(collection)
   COLLECTION_IMAGE_EXTENSIONS.map { |ext| "#{collection}#{ext}" }
 end
+
+GRAPHQL_BATCH_SIZE = 100
+
+def prefetch_all_collection_items!
+  return if @_prefetched
+
+  all_repos = []
+  all_users = []
+
+  collections.each do |collection|
+    items_for_collection(collection)&.each do |item|
+      if item.match?(USERNAME_AND_REPO_REGEX)
+        all_repos << item
+      elsif item.match?(USERNAME_REGEX)
+        all_users << item
+      end
+    end
+  end
+
+  all_repos.uniq!
+  all_users.uniq!
+
+  # Batch repos in chunks to stay within GraphQL query limits
+  all_repos.each_slice(GRAPHQL_BATCH_SIZE) do |batch|
+    cache_repos_exist_check!(batch)
+  end
+
+  # Batch users in chunks
+  all_users.each_slice(GRAPHQL_BATCH_SIZE) do |batch|
+    cache_users_exist_check!(batch)
+  end
+
+  # Check orgs for users not found
+  not_found_users = users_not_found_from(all_users)
+  not_found_users.each_slice(GRAPHQL_BATCH_SIZE) do |batch|
+    cache_orgs_exist_check!(batch)
+  end
+
+  @_prefetched = true
+end