fix(user_tools): handle zero-worker Databricks clusters

[WilliamK112]

Description

Normalize a missing Databricks executors field to an empty list in both the AWS and Azure cluster implementations. This lets zero-worker/single-node cluster initialization finish cleanly and reach the existing ClusterBase._verify_workers_exist validation instead of failing earlier with a TypeError while iterating over None.

The change deliberately preserves the current policy of rejecting driver-only clusters; adding support for that cluster shape remains out of scope.

The regression test covers both platforms using the existing Databricks fixtures with executors removed and num_workers set to zero, and verifies the intended "no worker nodes" error.

Closes #2143.

Validation

Run from user_tools on current dev (bcda54d):

• PYTHONPATH=src python -m pytest tests/spark_rapids_tools_ut/test_cluster.py -q — 9 passed
• Pylint on the two implementations and regression test — 10.00/10
• Flake8 on the two implementations and regression test
• py_compile on the two implementations and regression test
• git diff --check upstream/dev...HEAD

[greptile-apps]

Confidence Score: 5/5

The PR appears safe to merge, with both implementations preserving existing validation behavior while avoiding the earlier missing-executor iteration failure.

Summary

• Normalizes missing or null executor collections to an empty list for AWS and Azure Databricks clusters.
• Preserves the existing policy that driver-only clusters are rejected.
• Adds a parameterized regression test covering both platform implementations.

_{Reviews (1) · Last reviewed commit: "fix(user_tools): handle zero-worker Data..."}

[amahussein]

duplicate of #2148

[WilliamK112]

Thanks for pointing that out — I missed #2148 when checking for overlap. It covers the same zero-worker handling, so closing this duplicate makes sense.