plain-html — integration host

• 00-baseline-short — Baseline — short prose
• 01-baseline-long — Baseline — long prose
• 02-density-sparse — Sparse density — 5 terms in 5k words
• 03-density-extreme — Extreme density — 200 terms in 5k words
• 04-pathological-names — Pathological term names — overlap, punctuation, single-char, html-tag names
• 05-tables-heavy — Tables-heavy — terms inside cells, headers, captions
• 06-code-heavy — Code-heavy — fenced blocks + inline code + post-mount syntax highlighter
• 07-math-heavy — Math-heavy — inline + display math, KaTeX-rendered post-mount
• 08-nested-deep — Deep nesting — 6-level heading hierarchy + 5-level nested lists
• 09-i18n-german — German — long compound terms, sharp-s, umlauts
• 10-i18n-japanese — Japanese — CJK boundary detection, no-space prose
• 11-i18n-arabic-rtl — Arabic — right-to-left layout, dir=rtl host wrappers
• 12-openapi-derived — OpenAPI-derived — rendered from petstore-3.1.json via Glossa processor
• 13-arxiv-derived — arXiv-derived — "Attention Is All You Need" (Vaswani et al., 2017)
• 14-mega-doc — Mega doc — 500k words / 2k terms, perf stress
• 15-pathological-html — Pathological HTML — entities, smart quotes, ZWJ, emoji, control chars
• 16-rao-blackwell-theorem — Rao-Blackwell theorem — a faithful expository note
• 17-rao-blackwell-equations — Rao-Blackwell — the theorem in symbols