Better handle modeling predictions for items with *no* data

This doesn't generally happen, but did the other day when I rolled back
some of the database's SWF asset records but kept the items—and it was
a bit confusing that the homepage marked them as fully modeled!

app/models/item.rb

@@ -291,6 +291,10 @@ class Item < ApplicationRecord
       # This might just be a species-specific item. Let's be conservative in
       # our prediction, though we'll revise it if we see another body ID.
       compatible_body_ids
+    elsif compatible_body_ids.size == 0
+      # If somehow we have this item, but not any modeling data for it (weird!),
+      # consider it to fit all standard pet types until shown otherwise.
+      PetType.basic.distinct.pluck(:body_id).sort
     else
       # First, find our compatible pet types, then pair each body ID with its
       # color. (As an optimization, we omit standard colors, other than the

spec/models/item_spec.rb

@@ -29,11 +29,11 @@ RSpec.describe Item do
 		describe "an item without any modeling data" do
 			subject(:item) { items(:straw_hat) }
 
-			pending("is not fully modeled") { should_not be_predicted_fully_modeled }
+			it("is not fully modeled") { should_not be_predicted_fully_modeled }
 			it("has no compatible body IDs") do
 				expect(item.compatible_body_ids).to be_empty
 			end
-			pending("predicts all standard bodies are compatible") do
+			it("predicts all standard bodies are compatible") do
 				expect(item.predicted_missing_body_ids).to contain_exactly(1, 2, 3)
 			end
 		end