+1 to the goal! This migration looks like a big piece of work. Perhaps we could start with a couple of smaller speed-up fixes, like #76890 and #77192.
I also looked into the unit test goal. If users write unit tests as our docs suggest [1], those tests still rely on core (they use DagBag), so speeding up `import airflow` alone may not help them much. So I started a thread [2] about a testing module for Dag authors. One idea is to decouple it from core. Would be happy to hear your thoughts! Thanks, Andrew [1] https://airflow.apache.org/docs/apache-airflow/stable/best-practices.html#unit-tests [2] https://lists.apache.org/thread/wrpdzq8cc0mljlq63g73w5s7r9m2ly5m
