🔧 refactor: Display name logic in Parallel Responses (#11149)

- Removed the use of `getResponseSender` for computing display names in both `Agent.js` and `loadAddedAgent.js`.
- Implemented fallback logic to derive display names from `modelLabel`, `modelSpec.label`, and `endpointConfig.modelDisplayLabel`.
- Enhanced comments for clarity on the new display name resolution process.
- Updated `useChatFunctions` and `createDualMessageContent` to accommodate changes in sender logic, ensuring consistent handling of ephemeral agents across the application.

client/src/hooks/Chat/useChatFunctions.ts

@@ -18,6 +18,7 @@ import type {
   TMessage,
   TSubmission,
   TConversation,
+  TStartupConfig,
   TEndpointOption,
   TEndpointsConfig,
   EndpointSchemaKey,
@@ -169,6 +170,7 @@ export default function useChatFunctions({
     }
 
     const endpointsConfig = queryClient.getQueryData<TEndpointsConfig>([QueryKeys.endpoints]);
+    const startupConfig = queryClient.getQueryData<TStartupConfig>([QueryKeys.startupConfig]);
     const endpointType = getEndpointField(endpointsConfig, endpoint, 'type');
     const iconURL = conversation?.iconURL;
 
@@ -291,6 +293,7 @@ export default function useChatFunctions({
           conversation,
           addedConvo,
           endpointsConfig,
+          startupConfig?.modelSpecs?.list,
         );
       } else {
         initialResponse.content = [];