[FEA] Write test wrapper to run SQL queries via pyspark

[mythrocks]

There are Python integration tests that construct data sources and run SQL queries. Consider this window-function test:

https://github.com/NVIDIA/spark-rapids/blob/branch-0.2/integration_tests/src/main/python/window_function_test.py#L79
def test_window_aggs_for_ranges(data_gen):
    df = with_cpu_session(
        lambda spark : gen_df(spark, data_gen, length=2048))
    df.createOrReplaceTempView("window_agg_table")
    assert_gpu_and_cpu_are_equal_collect(
        lambda spark: spark.sql(
            'select '
            ' sum(c) over '
            '   (partition by a order by cast(b as timestamp) asc  '
            '       range between interval 1 day preceding and interval 1 day following) as sum_c_asc '
            # ...
            'from window_agg_table'))

It would be good to have a pytest wrapper that does the following:

1. Takes a test input data_gen, and a SQL string.
2. Constructs an input DataFrame, and a corresponding input table.
3. Runs the supplied SQL on CPU and GPU, and compares results.

Should be a tiny utility.