inspired by A. Sophie Blee-Goldman
So, that's exactly what we should do in this test. We need to
1. call KafkaStreams#cleanup before starting the application up the second time (ie when we turn optimizations on)
2. run the reset tool between stopping the original application and starting the new one
I think that 1. alone would technically be enough to stop this test from failing, but we should really be doing both.
the comment.1 is addressed by
KAFKA-10191 and this issue aims to address another one.