First, to get coverage on @jit functions, we have to disable the jit for the test run. This seems OK, since the correctness should not depend on whether it's been jitted, but it will make tests slower and I had to accommodate by reducing the runtime of some tests that were taking too long.
This PR adds coverage tooling, with some caveats.
First, to get coverage on
@jitfunctions, we have to disable the jit for the test run. This seems OK, since the correctness should not depend on whether it's been jitted, but it will make tests slower and I had to accommodate by reducing the runtime of some tests that were taking too long.