Refactor logic for parallel testing

[gerashegalov]

Reduce code duplication in run_pyspark_from_build.sh.

Borrowing logic tested in #1688

Tested with

# single threaded
TEST_PARALLEL=1 SPARK_HOME=~/spark-3.1.1-bin-hadoop3.2 integration_tests/run_pyspark_from_build.sh -k test_single_orderby

# parallel: 
TEST_PARALLEL=4 SPARK_HOME=~/spark-3.1.1-bin-hadoop3.2 integration_tests/run_pyspark_from_build.sh -k test_single_orderby

[gerashegalov]

build

[gerashegalov]

build

[gerashegalov]

build

[pxLi]

I saw this already been triggered not sure why trigger twice for the same commit. If the previous one is not required plz help abort to save resource, thx~

rapids_premerge-github 1253
rapids_premerge-github 1254

[gerashegalov]

@pxLi sorry, my browser UI probably did not refresh and I mistakenly thought my first build comment didn't go through. Now I see that I have duplicate build comments.