Test that .zenodo.json and paper.md agree on authors and affiliations

paper/paper.md and .zenodo.json both carry the author list, affiliations and ORCIDs, with nothing keeping them in step. They drifted once already: .zenodo.json was generated from paper.md on 08-25, #488 renumbered the affiliations on 08-26, and @mschwamb caught the mismatch reviewing #480.

It matters because .zenodo.json sets the author list on the archived record, which is what people cite, and nobody re-reads it before the DOI is minted.

Suggested test: parse both, check the same authors in the same order, matching ORCIDs, and each author resolving to the same institution names. Names, not index numbers — a renumbering can be correct in the affiliation list and still leave a stale index on one author, which is what happened to J. Moeyens in #488 ("8,13", where 8 had come to mean something else). An index comparison passes that.

Needs #480 to land first, since .zenodo.json is not on main yet.

After #480 lands, start by reading paper/paper.md and .zenodo.json and identify the repository's existing Python test entry point. Add a test that compares authors in order, ORCIDs, and resolved institution names rather than affiliation indices; done means the test catches stale or renumbered author affiliations.