XSL-FO: render captions as run-in paragraphs, not centered

[rbeezer]

In the XSL-FO → PDF route, a figure-like block's caption — a figure's caption, and the title of a table, listing, or list — was set in a text-align="center" block. A short caption survives this, but a long one renders as awkward centered, ragged, multi-line text. A caption is really a paragraph and should follow the same rules: a bold run-in heading (the type name and number) leading body-aligned prose.

This switches that block to the document's body alignment ($text-alignment — justify, or start when ragged-right), so the bold "Figure 5.3." / "Table 5.1." becomes a true run-in heading and the caption sets as ordinary prose. The paragraph following the figure already indents correctly via the p template's preceding-sibling::p rule, and a caption holds inline content, so there are no paragraphs within it to indent.

Testing

Sample article: the figure-long-caption test (Figure 10.31) now reads as a justified paragraph led by a bold "Figure 10.31." run-in heading, rather than a centered ragged block; short captions and the table/listing/list titles get the same flush run-in treatment. veraPDF --flavour ua1: 106 passed / 0 failed.

Commits

• XSL-FO: render captions as run-in paragraphs, not centered

Claude Opus 4.8, acting as a coding assistant for Rob Beezer

[rbeezer]

That Table of Contents commit was a mistake, it was meant to be part of #2937, and it should be gone now after a force-push.

[rbeezer]

While fixing #2937, I messed up #2966 here via a force-push. So the initial reported merge is gone, and I have re-done it as fb8d9b4, which is the most recent report.