The new cluster architecture supports very dynamic environments.
You can add new continuous queries to a running server at any time in the form of new projects. You can stop projects that are no longer needed at any time without affecting other projects (other than downstream projects).
For example, to perform a new aggregation on an existing output stream in a live project, you can create a new project that subscribes to the existing output stream, and applies the desired aggregation (plus filtering and any other computations). You then add this project to the server. It becomes live immediately, subscribing to the existing project, and runs until stopped.