Phase V8 — visual redesign closure, V-sequence complete
What changed
Deletes the orphaned pre-archetype placeholder scenes (EncounterIllustration/BossIllustration, BodyDiagram — zero live call sites), re-platforms EnemyIllustration onto the same archetype figure set (GenericFigure/TyrantFigure) the live combat HUD already uses so the prelude gallery and the fight screen agree on every foe's look, reconciles SVG_ASSET_SPEC.md's checklist to ground truth, runs a /critic-loop screenshot pass (25 screens) and fixes the one confirmed low-contrast finding (cutscene hint text, AXM.ash on near-black → AXM.bone), and adds docs/VISUAL_LANGUAGE.md documenting the finished Woodcut Codex system.
Why it changed
Closes the V-sequence (V1-V8) opened to reconcile the visual redesign; the orphaned scenes and split figure sets were debt left mid-migration, and V8 was scoped as the closure tick rather than a general UX pass — remaining findings are logged for /critique or /iterate to drain on their own loop.