lcov report has bad metrics when using sandbox

I tried producing a code coverage report using the lcov reporter. It works as long as I don't mess with the sandbox util.

I'm using the sandbox utility to replace 'require' and mock out modules used by the module under test.

I've been working really hard at trying different code coverage tools and different ways to mock.

Should I be mocking differently or is there a way we could fix the code coverage instrumentation?