Check for certificates that likely check the same paper without a shared work key
Source:R/utils_register_check.R
check_near_duplicate_works.Rdcodecheckers/register#150's flagship case (#133/#149, certificates 2024-017/2024-025): two certificates for the same paper, but one's `Paper reference` is a DOI and the other's is a preprint PDF URL, so [normalize_work_key()] gives them different (or missing) keys and they never land on the same `/works/` page. This is intentionally *not* resolved automatically - see the "Grouping" decision in the #150/#123 implementation plan (group by DOI only, report near-duplicates) - a title match is exactly the kind of ambiguous signal that should be reviewed by hand, not merged silently.