Tempest based benchmark scenario

Major goal is to reuse tons of already existing tempest scenarios in purpose of benchmarking.

So we should add something like TempestScenario(base.Scenarion) class.

And inside it couple of method:

def all(self)
   #just run all tests from tempest one by one
   pass

def set(set_name):
    # run one by one methods from set (set is hardcoded in inside Rally)
    pass

def random_test_from_set(set_name):
    # run random test from set with name set_name
    pass

def specific_regex(regex):
    # regexp that will find all benchmarks
    pass

def single_test(self, test_name):
    # run test with test_name
    pass

def list_of_test(self, test_names):
    # run one by one test from list of test_names
    pass

def random_test_from_list(self, test_names)
   # run random test from test_names
   pass