`JobQueueCluster` with local worker(s)

Thanks a lot for this fantastic library - it is really awesome!

I'd love to hear your opinion on the following use case. I have access to a SLURM cluster where I am not allowed (or I am at least discouraged) to run tasks such as a Jupyter server or a Dask scheduler on the login node, and the minimal partition size that I can request via a batch job is 32 cores.

I can start the Jupyter server and instantiate a `SLURMCluster` in a batch job, then scale up the cluster by adding resources via SLURM. However, the initial batch job where Jupyter and the Dask scheduler are running still occupies 32 cores - which is a bit wasteful. Right now I could "manually" create an additional local worker and connect it to the scheduler to fill the remaining allocated resources. But maybe it could be useful to have the option to add a local worker when instantiating a `JobQueueCluster`?