refactor(api): only fetch router models for providers with configured profiles. Fixes #4065

Add hasConfiguredProfile helper to filter model fetch candidates based on
listApiConfigMeta, preventing unnecessary API calls for providers without
configured profiles.

- Extract listApiConfigMeta from state to check configured providers
- Filter candidates to only fetch models for providers in use
- Make litellm and io-intelligence type properties optional
- Add apiKey to glama and vercel-ai-gateway candidate options
- Update tests to verify profile-based filtering behavior

Author: mcowger

src/api/providers/fetchers/modelCache.ts

@@ -96,7 +96,7 @@ export const getModels = async (options: GetModelsOptions): Promise<ModelRecord>
 				break
 			case "litellm":
 				// Type safety ensures apiKey and baseUrl are always provided for LiteLLM.
-				models = await getLiteLLMModels(options.apiKey, options.baseUrl)
+				models = await getLiteLLMModels(options.apiKey || "", options.baseUrl || "")
 				break
 			// kilocode_change start
 			case "kilocode": {