The regression was executed serially from the beginning
of construction, as time going, more tests are accumulated,
which consume much time on Concourse/Pipeline. So we modify
these tests and run them with parallel schedule.