fix(pptx): preserve advance rounding for Latin mathematical signs

[developer0hye]

Summary

Keep PowerPoint's 1/8pt nominal advance rounding when an otherwise Latin run contains ×, ÷, or ±. The marketing slide's final glyph endpoint moves from 1.073pt left of native PowerPoint to 0.018pt right. Whole-word shaping, spaces, and ASCII-hyphen break opportunities are preserved.

Related issue

Related: #1581. Cross-space kerning remains a separate issue (#1582).

Testing

• cargo test --locked --workspace --no-fail-fast: 3,489 passed, 15 ignored.
• Two portable regressions pass: six mathematical-sign width cases fail before the fix; wrapping coverage checks expressions, spaces, and hyphens.
• Fifteen native PowerPoint width/control probes preserve native line contents and agree in baseline within 0.12pt.
• cargo clippy --locked --workspace --all-targets -- -D warnings, cargo fmt --all --check, and git diff --check: passed.
• Seventeen-page layout audit passes at 0.5pt; searchable text matches (1,593 normalized codepoints and identical codepoint-class census).
• All seventeen strict cluster reports pass with zero material 5% fuzz clusters.
• Documentation freshness review: PASS.

Visual impact

• No rendered PDF change
• Rendered PDF change or visual evidence added

Visual audit

• Issue: PPTX: multiplication sign disables advance-grid rounding for the entire Latin run #1581
• Fixture: tests/golden_mocks/business/sources/pptx/08_marketing_report_en.pptx, followed by mathematical-sign and text-box-width probes. Mapping, reproduction, and hashes: assets/bugfixes/issue-1581/README.md.
• Page(s): 1-17
• Renderer and DPI: pdftoppm, 150 DPI; native PowerPoint 16.112.3 GT
• Evidence mode: fix
• Layout audit report: assets/bugfixes/issue-1581/layout-audit.json
• Render cluster reports: assets/bugfixes/issue-1581/render-clusters-page-1.json, assets/bugfixes/issue-1581/render-clusters-page-2.json, assets/bugfixes/issue-1581/render-clusters-page-3.json, assets/bugfixes/issue-1581/render-clusters-page-4.json, assets/bugfixes/issue-1581/render-clusters-page-5.json, assets/bugfixes/issue-1581/render-clusters-page-6.json, assets/bugfixes/issue-1581/render-clusters-page-7.json, assets/bugfixes/issue-1581/render-clusters-page-8.json, assets/bugfixes/issue-1581/render-clusters-page-9.json, assets/bugfixes/issue-1581/render-clusters-page-10.json, assets/bugfixes/issue-1581/render-clusters-page-11.json, assets/bugfixes/issue-1581/render-clusters-page-12.json, assets/bugfixes/issue-1581/render-clusters-page-13.json, assets/bugfixes/issue-1581/render-clusters-page-14.json, assets/bugfixes/issue-1581/render-clusters-page-15.json, assets/bugfixes/issue-1581/render-clusters-page-16.json, assets/bugfixes/issue-1581/render-clusters-page-17.json
• Reference exporter differences: None
• Fine-detail threshold: 0.5pt
• Layout audit page count: Pass
• Layout audit text flow: Pass
• Layout audit visible fills: Pass
• Layout audit rectangle geometry: Pass
• Layout audit large shifts: Pass
• Layout audit fine shifts: Pass
• New follow-up issues found in this audit: None
• Model vision findings: Full-page context, full-scale content crops, and diffs show the original bullet's accumulating suffix drift gone. Mathematical glyph shapes, bold red headings, regular body text, bullet indents, and native numeric-expression and hyphen wrapping remain intact. Original slides 1 and 2 have identical decoded GT/output/diff pixels to the previously inspected PPTX: wrapped continuation baselines retain unsnapped fractional line offsets #1584 comparison, including the chart's thin axes, ticks, bar edges, and legend outline. Sparse heading/body glyph-edge rasterization fragments are bounded observations; there are no material diff clusters or remaining converter deviations in these seventeen compared pages.
• GT: assets/bugfixes/issue-1581/gt.jpg
• Before: assets/bugfixes/issue-1581/before.jpg
• After: assets/bugfixes/issue-1581/after.jpg
• Native: None
• Compare: assets/bugfixes/issue-1581/compare.jpg

Visual comparison

GT	Before	After

Required inspection

• Rendered all evidence at 150 DPI or higher
• Stored progressive JPEG quality 86 assets with metadata stripped
• Used Codex/Claude vision to inspect the full GT/output pages, diff, and matched crops
• Inspected matched region crops at full resolution
• Ran compare_layout.py --audit --fine-shift PT and dispositioned every fine/large text-instance shift, rectangle geometry deviation, painted-text visibility mismatch, and visible-fill occlusion
• Ran compare_render.py --cluster-report PATH --strict-clusters and dispositioned every material 5% fuzz diff cluster by explicit ID
• Inventoried hairlines and border dash styles
• Inventoried font weight, italic, and underline emphasis

Deviation audit

Check	Result
Page count/order	Matches GT: seventeen pages in documented fixture/probe order.
Element presence	Matches GT: original titles, bodies, bullets, chart, and probe text remain present.
Position/size	Fixed: the original endpoint drift decreases from -1.073pt to +0.018pt; all layout gates pass.
Rotation/flip	No deviation observed.
Fill	Matches GT: red/white cover, chart, and white text pages retain native fills.
Stroke/border	Matches GT: chart axes, ticks, bar edges, and legend outlines are unchanged.
Shape outline geometry	Matches GT: no material matched rectangle or outline deviation.
Text content	Matches GT: searchable content and codepoint-class census match.
Font family/weight/style	Matches GT: Arial bodies, bold headings, and mathematical glyph shapes remain intact.
Text color	Matches GT.
Alignment	Matches GT: left alignment, bullet indentation, and continuations are preserved.
Line/paragraph spacing	Matches GT: expression, space, and hyphen wraps match all fifteen native width/control cases.
Clipping/overflow	No deviation observed.

Checklist

• Commits include a Signed-off-by line
• PR scope contains one root cause
• Remaining converter or harness deviations each reference an open issue