Add a workflow example that uses multi-node Databricks and Dask (ideally also using dask-deltatable)

[jacobtomlinson]

### Prep tasks
- [x] Choose a workload (xgboost training, cuml training)
- [x] Put a dataset into Delta Lake
- [x] Figure out how to read with dask-deltatable into a dask-cudf dataframe

Workflow structure

Option 1 (goal):

• Start MNMG Dask/RAPIDS cluster on Databricks (what we have in the docs now)
• Read dataset from Delta Lake with dask-deltatable into a dask-cudf dataframe
• Do some preprocessing
• Train a model
• Summary

Option 2 (stretch goal):

• Start MNMG Dask/RAPIDS cluster on Databricks (what we have in the docs now)
• Generate data somehow in a GPU intensive way (custreamz)
• Append records to an existing Delta Lake table

Related links:

• https://pypi.org/project/dask-deltatable/
• https://blog.dask.org/2023/02/02/easy-cpu-gpu
• https://learn.microsoft.com/en-us/azure/databricks/delta/tutorial
• https://delta.io/blog/2022-10-25-create-delta-lake-tables/
• https://medium.com/rapids-ai/gpu-accelerated-stream-processing-with-rapids-f2b725696a61