ci: Add rerun failures package for flaky tests

[CagriYonca]

I'm taking the initiative and proposing the pytest-rerunfailures package for flaky tests. Only thing we should do is to add the line just before flaky tests:

    @pytest.mark.flaky(reruns=3, reruns_delay=2)   <-- ONLY THIS
    def test_flaky_test(self):
        ...

[CagriYonca]

Grpcio fails randomly:

• FAILED tests/frameworks/test_grpcio.py::TestGRPCIO::test_server_error

    def test_server_error(self) -> None:
        response = None
        with tracer.start_as_current_span("test"):
            try:
                response = self.server_stub.OneQuestionOneErrorResponse(
                    stan_pb2.QuestionRequest(question="Do u error?")
                )
            except Exception:
                pass
    
        assert not get_current_span().is_recording()
        assert not response
    
        spans = self.recorder.queued_spans()
>       assert len(spans) == 4
E       assert 3 == 4

https://app.circleci.com/pipelines/github/instana/python-sensor/3997/workflows/8223d7a0-aaa6-4c73-b322-6ea4c97f4ae5/jobs/30757/parallel-runs/0/steps/0-105

[CagriYonca]

Kafka fails randomly:

• tests/clients/kafka/test_confluent_kafka.py::TestConfluentKafka::test_confluent_kafka_poll_root_exit

    def test_confluent_kafka_poll_root_exit(self) -> None:
        agent.options.allow_exit_as_root = True
    
        # Produce some events
        self.producer.produce(testenv["kafka_topic"], b"raw_bytes1")
        self.producer.flush()
    
        # Consume the events
        consumer_config = self.kafka_config.copy()
        consumer_config["group.id"] = "my-group"
        consumer_config["auto.offset.reset"] = "earliest"
    
        consumer = Consumer(consumer_config)
        consumer.subscribe([testenv["kafka_topic"]])
    
        msg = consumer.poll(timeout=30)  # noqa: F841
    
        consumer.close()
    
        spans = self.recorder.queued_spans()
        assert len(spans) == 2
    
        producer_span = get_first_span_by_filter(
            spans,
            lambda span: span.n == "kafka"
            and span.data["kafka"]["access"] == "produce"
            and span.data["kafka"]["service"] == "span-topic",
        )
    
        poll_span = get_first_span_by_filter(
            spans,
            lambda span: span.n == "kafka"
            and span.data["kafka"]["access"] == "poll"
            and span.data["kafka"]["service"] == "span-topic",
        )
    
        # Same traceId
>       assert producer_span.t == poll_span.t
E       AssertionError: assert 6259974530922409176 == 3135373471331694061

https://app.circleci.com/pipelines/github/instana/python-sensor/4060/workflows/483538f3-52bc-458e-af45-87b60c6511da/jobs/31584/parallel-runs/0/steps/0-105

[CagriYonca]

Tornado fails randomly:

• tests/frameworks/test_tornado_server.py::TestTornadoServer::test_get_with_params_to_scrub

    def test_get_with_params_to_scrub(self) -> None:
        async def test():
            with tracer.start_as_current_span("test"):
                async with aiohttp.ClientSession() as session:
                    return await self.fetch(session, testenv["tornado_server"], params={"secret": "yeah"})
    
        response = tornado.ioloop.IOLoop.current().run_sync(test)
    
        spans = self.recorder.queued_spans()
>       assert len(spans) == 3
E       AssertionError: assert 2 == 3

https://app.circleci.com/pipelines/github/instana/python-sensor/4071/workflows/f52e3efe-42c0-4972-8a99-291b961743e1/jobs/31732/parallel-runs/0/steps/0-105