Functionality for returning "known unknowns" was implemented in a fairly nonstandard way by bubbling unknowns with errors in order to keep compatibility with the Syft 1.0 APIs.

Assuming we end up with an approach similar to what some discussions revolved around: tasks and catalogers are passed an SBOM object to populate, which has thread-safe functions to perform common tasks like: adding packages, adding relationships, and adding unknowns. This is a placeholder for the latter function: adding unknowns.