refactor(landing): memoize dashboard cards, lazy-load below-fold sections, shrink logo

[Codesmith313]

Summary

Three perf issues, all assigned to me:

• perf: memoize expensive balance and payment list renders #479 — TopAssets, ActivityHighlights, and QuickConvert (the home dashboard's balance/activity sections) now wrapped in React.memo, so an unrelated state update on the home page no longer re-renders the whole dashboard. Added a test confirming each is actually memo-wrapped.
• perf: lazy-load landing page sections to improve initial LCP #477 — The landing page's use-cases, FAQ, and footer sections are now behind next/dynamic with skeleton loading fallbacks, so they're split out of the page's initial render path instead of loading eagerly alongside the hero.
• perf: replace PNG logos with optimised SVG/WebP using next/image #478 — public/logo.png (313 KB, turned out to be a fully unreferenced raw Figma export — nothing in the app links to it) re-encoded as an 800px-wide WebP at ~7.5 KB. Added sizes to the hero image next to its existing priority. There were no raw <img> tags anywhere in app/components to convert — landing already uses next/image throughout.

Closes #477
Closes #478
Closes #479

A blocking bug found along the way

components/session-provider.tsx didn't compile on dev — a bad merge between the Freighter-auth PR and the session-persistence PR left duplicate imports, a duplicate local Session type shadowing the one imported from @/lib/api, and two interleaved signOut implementations (the second literally opening inside the first's unclosed body). This is what npm test (the pre-push hook's full suite) has to be able to compile, so it needed fixing here too. Reconciled it to the cookie-based api.getSession()/api.logout() restoration path (the direction issue #582 already points toward, away from storing the JWT in localStorage) plus the Freighter sign-in flow, dropping the now-redundant local Session interface and localStorage fallback in favor of the type @/lib/api already exports.

Notes

• next/dynamic's ssr: false isn't usable from a Server Component, so these three sections stay server-rendered — the benefit here is streaming (the hero + above-fold content isn't blocked on rendering the below-fold sections) rather than a client-JS-bundle reduction, since none of UseCases/Faq/SiteFooter ship client JS in the first place. I didn't set up a Lighthouse CI run to produce a hard LCP percentage for this environment — that needs a deployed/built target — but streaming below-fold Server Components is a standard technique for improving TTFB/LCP.
• npm run build currently fails on dev for reasons unrelated to this PR — app/(app)/transactions/page.tsx and app/request/[id]/page.tsx both have the same "bad merge" pattern as session-provider.tsx (duplicate imports, dozens of now-undefined names), and lib/api.ts references an undefined Sep24Interactive type. None of these are touched by this PR.

Test plan

• npx jest — 35/35 passing, including the pre-push hook's own full run
• npx eslint on every changed file — clean
• npx tsc --noEmit — no new errors introduced (pre-existing errors are all in files this PR doesn't touch)
• Visually diffed logo.webp against the original logo.png — pixel-identical at 800px, well under the 10 KB target

[drips-wave]

@Codesmith313 Great news! 🎉 Based on an automated assessment of this PR, the linked Wave issue(s) no longer count against your application limits.

You can now already apply to more issues while waiting for a review of this PR. Keep up the great work! 🚀

Learn more about application limits

[vercel]

@Codesmith313 is attempting to deploy a commit to the kelly musk's projects Team on Vercel.

A member of the Team first needs to authorize it.