This is especially important for topics that contain statistical data, such as airFlow and airTurbulence. It would be best to check that the statistics were correctly computed.

Also make the following changes:

• Test the sensorState event.
• Use write_and_wait instead of assert_next_severity in most or all tests, to improve reliability. Remove assert_next_severity if practical, or add a flush_severities function so we can use it occasionally without worrying about whether severities have accumulated. Or, better yet, consider ways to write a lot of data and then test that it was all received – it would speed up tests and also do a better job of detecting certain race conditions. That could consist of counting messages that are processed and waiting for the number to get to a certain point. It can't consist of clearing events and waiting for them to be set, since that will lead to race conditions.