When an inline site has a starting code offset of 0, we sometimes
don't emit the starting offset.
This patch removes the case that emits a BinaryAnnotationOpCode::ChangeLineOffset when the code offset is 0. These cases will just go to the next case that emits a BinaryAnnotationOpCode::ChangeCodeOffsetAndLineOffset.
I'd probably suggest a simpler (it's probably about the same in IR, but simpler in terms of language features, names, etc) test case for inlining, something like:
If that's adequate.