It probably would. But not everyone will want column-level breakpoints? This is part of what I’m getting at with it being a UX question rather than correctness question.
Moving is_stmt up to the first instruction in the atom containing the KI, rather than on the KI itself, might help address this point. The scope of the atom could vary depending on whether we had column-level breakpoints enabled, so this movement would or would not cross a column change depending.