Add a commented out WATSON_STT_MODEL variable to the docker compose yml sample

[jfmartinez]

Currently in our docker-compose files, we don't specify the WATSON_STT_MODEL. It'd be easier for customers to be able to just uncomment it and specify it as needed. Additionally, it'd be better if we add a link to the list of Watson Speech to Text models available.