Some thoughts I've had along these lines (but for Rust) - Open issues in the review for compiler / lint / format errors - Gather coverage per test and use that

Some thoughts I've had along these lines (but for Rust) - Open issues in the review for compiler / lint / format errors - Gather coverage per test and use that to determine which tests are needed for a given change