BibMedEd v0.3.0 — sample workflow and contributor improvements

Added

• Added a deterministic, network-free synthetic sample workflow (POST /api/projects/sample) with 12 records, meaningful author/country/keyword/citation networks, screening state, and a reproducible methodology trace. Repeated requests reopen the same sample instead of creating unbounded copies.
• Added the bibmeded search --dry-run CLI for estimating an upstream result count without fetching or persisting records (#17).
• Added full bibmeded search execution through the existing database and Celery pipeline, including year and result-limit options (#52).
• Added Lens.org Scholarly as the fifth built-in source, with token-based configuration, pagination, record mapping, fixture-based tests, and empty-result handling (#14).

Changed

• Improved first-run usability with a homepage shortcut that creates the sample project and opens its analysis dashboard immediately.
• Updated the homepage, documentation, and adapter registry to reflect all five built-in sources.
• Clarified how newcomers claim work, how adapter auto-discovery works, and what a complete adapter contribution includes.

Fixed

• Replaced CLI tracebacks for unknown sources and network failures with actionable stderr messages and non-zero exit codes (#50).
• Prevented full CLI searches from polling forever when adapter construction fails.
• Handled empty Lens.org result pages without raising an indexing error.

Security

• Updated Next.js and its matching lint configuration from 16.2.9 to 16.3.1, then refreshed transitive dependencies so npm audit reports zero known vulnerabilities.

Community

• Thank you to @BaygeldiAza for the dry-run and full CLI search contributions (#46, #52).
• Thank you to @landon-personal for improving CLI source and network error handling (#50).