Use when auditing a codebase, test suite, benchmark, or plan in depth before a release, after a big milestone, or when a single reviewer keeps missing whole classes of problems.
