i also noticed while running the extra integration test i wrote (which i selected with

-k

) that the test result seemed to be cached even after i changed one of the source files (no pytest output, immediate completion), and i had to invalidate the target to get the test to run again (i know i should be using cache-ignore instead). i'm not as familiar as i should be with (1) whether we're caching python tests yet (2) how that works in any way, but why might that happen? is it because integration tests don't necessarily import the source files that affect their output? or some other reason?