We use a Retell AI voice agent workflow (question nodes, extract variable nodes, function nodes) in a Next.js app with RetellWebClient for web calls. Currently, users can only respond via voice. We want to allow users to type text answers in the same conversation while keeping the same voice agent workflow. Questions: How do we send a text message to an active web call? What API method/endpoint should we use? How do we get the callId or conversationId from an active RetellWebClient web call? Can a voice agent workflow accept text input mid-conversation, and will the extract variable nodes and question flow continue normally? Please provide API documentation or code examples. Thank you.