Documented example on dask-gateway client configured scheduler memory limits

[consideRatio]

I'm quickly opening this ticket as I'd like to investigate and provide an example on exposing options for end users via the dask-gateway client to configure a scheduler's cpu and memory request/limits for the Kubernetes backend.

[caio-eiq]

@consideRatio do you have any example handy for now so I can take a look? Also, does the Kubernetes backend allow me to pass a custom namespace in case I need to launch schedulers & workers in a custom namespace rather than the default where dask-gateway has been deployed to? I'd like to see an example in code rather than making the changes in the Helm Chart.