Refactor Dask cluster boostrapping

docs/source/quickstart.md

@@ -15,17 +15,15 @@ reduce the output or write it to disk after shuffling.
 ```python
 import dask.distributed
 import dask.dataframe as dd
+from dask_cuda import LocalCUDACluster
 
 from rapidsmp.examples.dask import dask_cudf_shuffle
-from rapidsmp.integrations.dask import LocalRMPCluster, bootstrap_dask_cluster
 
 
 df = dask.datasets.timeseries().reset_index(drop=True).to_backend("cudf")
 
-with LocalRMPCluster() as cluster:
+with LocalCUDACluster() as cluster:
     with dask.distributed.Client(cluster) as client:
-        bootstrap_dask_cluster(client)
-
         shuffled = dask_cudf_shuffle(df, shuffle_on=["name"])
 
         # collect the results in memory.