[BUG] Failed test against Spark 3.2.0 using a single-shim jar on Databricks

[razajafri]

Describe the bug
Databricks tests will try to run the tests against spark-3.2.0 even if we build a single shim jar. e.g.

Steps/Code to reproduce bug
WITH_DEFAULT_UPSTREAM_SHIM=0 ./jenkins/databricks/build.sh
./jenkins/databricks/test.sh will quietly exit with the following output

+ /home/ubuntu/spark-3.2.0-bin-hadoop3.2/bin/spark-shell --master 'local-cluster[1,2,1024]' --conf spark.plugins=com.nvidia.spark.SQLPlugin --conf spark.deploy.maxExecutorRetries=0 --conf spark.shuffle.manager=com.nvidia.spark.rapids.spark320.RapidsShuffleManager --driver-class-path /home/ubuntu/spark-rapids/dist/target/rapids-4-spark_2.12-24.10.0-SNAPSHOT-cuda11.jar:/home/ubuntu/spark-rapids/integration_tests/target/rapids-4-spark-integration-tests_2.12-24.10.0-SNAPSHOT-spark341db.jar:/home/ubuntu/spark-rapids/integration_tests/target/dependency/parquet-hadoop-1.12.0-tests.jar --conf spark.executor.extraClassPath=/home/ubuntu/spark-rapids/dist/target/rapids-4-spark_2.12-24.10.0-SNAPSHOT-cuda11.jar:/home/ubuntu/spark-rapids/integration_tests/target/rapids-4-spark-integration-tests_2.12-24.10.0-SNAPSHOT-spark341db.jar:/home/ubuntu/spark-rapids/integration_tests/target/dependency/parquet-hadoop-1.12.0-tests.jar
+ grep -F 'res0: Array[org.apache.spark.sql.Row] = Array([4950])'

because it tries to run tests against spark-3.2.0 which we didn't build for.

Expected behavior
We should detect if we should run the spark-3.2.0 tests