Create public-domain test infobase

We currently use a private test infobase (the folio help file), which prevents our CI servers from passing, and makes it difficult for others from contributing to the project. It's also not terribly complete.
### We want to create a test infobase that
- Contains only public-domain content
- Has a consistent appearance, so that the corresponding (exported) HTML can be visually compared with relative ease, and issues quickly spotted.
- Is small (preferably 500kb or less). This is for speed, which matters when you run tests millions of times.
- Exercises all modern (and legacy) folio flat file features.
### Exercises
- Tables (including header rows, columns, cells, spacing, padding, borders, colors, and spans)
- Tab stops (including left, center, right, dot/dash/alternating leaders, and custom series)
- Fonts, spacing, margins, justification
- Cross-infobase queries and links (this will require a 2nd infobase)
- Headings
- Bookmarks, jump destinations
- Popups, named popups, notes
- Fields, including overlapping field applications
- Hidden text
- Underlining, strikeout, italics
- Levels, hierarchies, including skipping levels
- Objects and object links (including all image formats - bitmap, png, gif, jpeg)
- Data links (pdfs, spreadsheets, etc)
- Program links
- Highlighters, link styles, and character styles
- Web links
- Query links (including level, field, and group queries). Text indexing rules (we want to verify that we mimic the Folio search algorithm 100% when resolving these).
- Groups
### And all the esoteric, undocumented (or supposedly invalid) flat file codes that we've encountered over the years, like:
- TS (empty tab stop with no parameters)
- [add here, or comment]

Start by examining the current private test infobase, identified as the folio help file, and the CI or test setup that consumes it. Define a small public-domain replacement plus a second infobase for cross-links, keeping the main file preferably under 500 KB. Done means the fixture exercises the listed flat-file features and legacy codes while supporting consistent exported-HTML comparison.

Written by the indexing model from the issue text.