FIX: Prevent LLM enumerator from erroring when spam enabled (#1045)

This PR fixes an issue where LLM enumerator would error out when `SiteSetting.ai_spam_detection = true` but there was no `AiModerationSetting.spam` present.

Typically, we add an `LlmDependencyValidator` for the setting itself, however, since Spam is unique in that it has it's model set in `AiModerationSetting` instead of a `SiteSetting`, we'll add a simple check here to prevent erroring out.

Author: keegangeorge

lib/configuration/llm_enumerator.rb

@@ -38,7 +38,7 @@ def self.global_usage
           rval[model_id] << { type: :ai_embeddings_semantic_search }
         end
 
-        if SiteSetting.ai_spam_detection_enabled
+        if SiteSetting.ai_spam_detection_enabled && AiModerationSetting.spam.present?
           model_id = AiModerationSetting.spam[:llm_model_id]
           rval[model_id] << { type: :ai_spam }
         end

spec/configuration/llm_enumerator_spec.rb

@@ -0,0 +1,21 @@
+# frozen_string_literal: true
+
+RSpec.describe DiscourseAi::Configuration::LlmEnumerator do
+  fab!(:fake_model)
+
+  describe "#global_usage" do
+    before do
+      SiteSetting.ai_helper_model = "custom:#{fake_model.id}"
+      SiteSetting.ai_helper_enabled = true
+    end
+
+    it "returns a hash of Llm models in use globally" do
+      expect(described_class.global_usage).to eq(fake_model.id => [{ type: :ai_helper }])
+    end
+
+    it "doesn't error on spam when spam detection is enabled but moderation setting is missing" do
+      SiteSetting.ai_spam_detection_enabled = true
+      expect { described_class.global_usage }.not_to raise_error
+    end
+  end
+end