Integrate shared/ui into widgets/version-footer (PRD-018 SC-2 batch 1/3)

[fedorovvvv]

Sub-issue of #81. Batch 1/3 of the shared/ui follow-up integration deferred from #68.

Refs: PRD-018 SC-2 · RFC-016 · EVID-022 · closing EvidencePack EVID-023.

Closed by PR #85 (commit 434e179 for this batch).

Why

widgets/version-footer/ is the smallest integration surface and is a good first batch — it exercises ~half of the visual-atom primitives (Badge, Button, Progress, Spinner, Alert) on a single, well-contained widget. Landing it first establishes the migration pattern (atom-by-atom replacement, no behavioural change) before the larger mosaic surface in batch 2.

What landed

widgets/version-footer/ui/UpdateButton.svelte (commit 434e179):

• Inline <button class="update-btn"> → <Button variant="secondary" size="sm">
• Outer .update-btn-pos wrapper kept for fixed positioning (50/50 split: positioning stays inline, visual atom is now a primitive).
• Inner glow dot + UPDATE label + arrow + version chip kept as decorative spans.

UpdateDialog.svelte and VersionFooter.svelte reviewed and intentionally not modified — UpdateDialog already uses Dialog/Button/Code primitives; VersionFooter is fixed-positioned chrome (single bordered pill spanning two text segments), not an atom-mappable surface. Documented in PR description.

Acceptance criteria

• grep -RIn '<style>' template/src/widgets/version-footer shows only layout/motion styles, not atom styles. Verified post-merge: 3 style blocks remain (UpdateButton/UpdateDialog/VersionFooter), zero atom selectors (grep -E '\.(btn|badge|alert|progress|spinner)\s*\{' returns nothing).
• No new ad-hoc reimplementation of an atom that already exists in @/shared/ui.
• npm run check passes (0 errors, 0 warnings, 1039 files).
• Visual smoke in Chrome on / (footer visible) — 0 console errors, dual-theme parity (light + dark verified via MCP).
• Bundle delta inside PRD-018 NFR-002 budget. Pre-merge measurement against develop: net diff is −7 LOC in this widget (33 lines removed, 26 added), so primitive use does not increase JS by any measurable amount. Pre-Create and promote components #68 cumulative baseline not re-measured (would require checkout-and-build of the old commit), but PR-internal effect is neutral/negative.
• Diff is reviewable in one pass — UpdateButton.svelte single-file change, well under the 300-LOC target.

[fedorovvvv]

Closed by PR #85 (commit 434e179). All AC verified post-merge. EvidencePack: EVID-023.