Include pipeline viz data as figures in published manuscripts

[rec3141]

Summary

Generated manuscripts currently have no figures. The pipeline's viz preprocessor already creates chart-ready JSON data at viz/data/*.json (overview, mags, contigs, genes, taxonomy, coverage, etc.) — these should be embedded in the published paper.

Current state

• ai/figure_generator.py exists but duplicates chart logic that the viz preprocessor already handles
• Viz JSON data is available in the sqsh mount at viz/data/
• The danaSeq viz SPA and Marimo notebooks already render these charts
• Don't want to maintain multiple copies of figure generation scripts

Approach

• Copy viz JSON files into paper repo at results/figures/
• Embed as interactive Plotly HTML figures in the Quarto manuscript (HTML output)
• Generate static PNG/SVG for PDF output using kaleido or similar
• Single source of truth: pipeline viz preprocessor generates the data, paper just renders it

Available viz data (per submission)

• overview.json — assembly_size, n50, n_contigs, n_mags, hq/mq/lq counts
• mags.json — per-MAG completeness, contamination, size, taxonomy
• contig_lengths.json — length distribution histogram, coverage scatter
• genes.json — per-contig gene annotations
• Plus taxonomy, coverage, KEGG, MGE data when available

Priority: P2