Support Grouping Sets in DJ SQL

[shangyian]

For example

dj.sql(
    metrics = "my_metric",
    dimensions=[
        ["dim1"],
        ["dim1", "dim2"]
    ],
    filters=[],
    measures=False,
)

should return

SELECT dim1, dim2, metric_expression
FROM transform_node
GROUP BY
  GROUPING SETS (
    (dim1),
    (dim1, dim2)
  )

[anhqle]

To support this, for the Python client, it's pretty straightforward to convert the dimensions arg from a list to a nested list.

It's more unclear how the API endpoint should evolve :) I'll leave that up to you. Thank you!