[dagster-databricks] Fix setting databricks cluster node configuration (

#20000)

## Summary & Motivation
When submitting a Databricks Job, the DatabricksJobRunner is not setting
the cluster's `driver_instance_pool_id` node configuration correctly. As
a result, it is not possible to launch a Databricks Job from Dagster to
run on Databricks instance pools. This sets the cluster's node
configuration in accordance with the Databricks Jobs API to allow
launching Databricks Jobs on instance pools.

<img width="1522" alt="Screenshot 2024-02-28 at 8 49 36 PM"
src="https://github.com/dagster-io/dagster/assets/23409221/c1f806d0-2f3c-4364-8fb0-9e14d03640c6">

The code is currently attempting to access `driver_node_type_id` from
`cluster.new.nodes`. However, the step launcher's `run_config` spec
states that the `cluster.new.nodes` field expects `driver_node_type_id `
and `node_type_id ` to be nested within an object called `node_types`.
The fields `driver_instance_pool_id` and `instance_pool_id` are
first-class properties alongside `node_types` (see `run_config` spec
[here](https://github.com/dagster-io/dagster/blob/2cabe8733cb517c8caaa21e0b323f46b944ef3ef/python_modules/libraries/dagster-databricks/dagster_databricks/configs.py#L362)).


## How I Tested These Changes
I launched Dagster runs in a Databricks workspace from local Dagster
deployment.

python_modules/libraries/dagster-databricks/dagster_databricks/databricks.py

@@ -564,7 +564,7 @@ def submit_run(self, run_config: Mapping[str, Any], task: Mapping[str, Any]) ->
                 new_cluster["instance_pool_id"] = nodes["instance_pool_id"]
 
                 if "driver_instance_pool_id" in nodes:
-                    new_cluster["driver_node_type_id"] = nodes["driver_node_type_id"]
+                    new_cluster["driver_instance_pool_id"] = nodes["driver_instance_pool_id"]
             else:
                 node_types = nodes["node_types"]
                 new_cluster["node_type_id"] = node_types["node_type_id"]