It gets tossed onto the pipeline queue when a request is made to render the report. There's currently no scheduling component to it.

It was originally added as a way to render potentially long running R reports asynchronously and was envisioned to be used in conjunction with the R report caching checkbox (which is only available in studies).