v0.13.0 - Content-preserving builds, smaller repairs

v0.13.0 - Content-preserving builds, smaller repairs

The model composes the story and design. The skill preserves editable source,
fits the evidence, and reports what needs attention.

• Shared layout allocation reserves room for captions, bottom callouts, and
  footers. Content and readable type take priority over decorative framing.
• Numbered action panels keep their authored reading order, aligned peer boxes,
  and consistent padding. Editorial evidence gets clearer left insets and
  separation between figures, labels, and supporting text.
• Native chart labels retain fractional values by default, with explicit number
  formats respected. Compact category labels reflow without dropping words;
  anonymous decision bullets no longer receive guessed owner/trigger labels.
• Normal QA checks that mapped source text, table caveats, chart values, and
  axis units survive into the PowerPoint. Missing evidence produces a source
  location for repair; the check does not establish factual truth.
• Briefs can point to a few original local slide examples selected by purpose
  and content shape. The examples guide composition without imposing a slide
  sequence or loading the full corpus. GPT-6.1 Sol is a supported profile alias;
  one capable model can still complete the workflow.
• Quick finalization reuses verified unchanged renders automatically. Source
  checks and visual approval stay fresh. --no-render-cache forces a cold render.

The release examples use three distinct synthetic topics: a clean lab report,
an editorial demand story, and an operations decision. One earlier Luna-authored
source is retained as a compatibility example. These are authored design studies,
not a controlled claim that one model is better than another.

The GitHub skill and Codex plugin share the same runtime. MIT licensing is unchanged.