Add paper ingestion quality gate

[pine-harbor]

/claim #13

Summary

Adds paper-ingestion-quality-gate, a focused intake guard for the AI Paper Summarizer in SCIBASE's AI-Assisted Research Tools MVP. It checks whether an uploaded or repository-linked paper has been extracted cleanly enough before the summarizer generates key findings, implications, or next steps.

This is intentionally narrow: it does not score the generated summary, validate citation context, check journal venue quality, assess statistics, run reporting guidelines, detect plagiarism, review disclosures, or audit figure/table claim support. It sits earlier in the workflow: before summarization, when broken PDF/OCR/table/reference extraction can quietly poison downstream AI output.

What It Covers

• Dropped pages between the source PDF and extracted text.
• Required scientific sections: abstract, methods, results, and discussion.
• OCR confidence and unrecognized glyph noise.
• Table extraction completeness before numeric key findings are generated.
• Figure caption extraction before figure-derived summary bullets are allowed.
• Reference parsing coverage before citation-based implications are generated.
• Equation extraction coverage for mathematical or methods summaries.
• Required extraction provenance for text, sections, tables, figures, references, and digests.
• Unsafe key-finding generation while intake findings are unresolved.

Included Artifacts

• Dependency-free Node module: paper-ingestion-quality-gate/src/index.js
• Synthetic intake packets: paper-ingestion-quality-gate/data/sample_ingestion_packets.json
• Tests: paper-ingestion-quality-gate/test/paper-ingestion-quality-gate.test.js
• Demo/report generator: paper-ingestion-quality-gate/scripts/demo.js
• Generated reviewer packet, JSON summary, SVG summary, and MP4 slate in paper-ingestion-quality-gate/reports/

Validation

• npm.cmd test: 8 tests passed.
• npm.cmd run check: syntax checks passed.
• npm.cmd run demo: generated 4-packet report with SUMMARIZE: 1, REVIEW: 1, HOLD: 2.
• npm.cmd run video: generated reports/demo.mp4.
• git diff --check: passed.
• ffprobe: H.264, 1280x720, 8.000 seconds, 30 fps, 240 frames.

The MP4 is a generated FFmpeg color-source/drawtext slate, not a desktop or browser recording. It contains no private data.