[DRAFT] import data src > local app creation

[callumflack]

Summary

• implement the first App Quickstart slice as a source-scoped Create app flow
• add the supporting spec / invariants / implementation-plan docs for the redesigned quickstart contract
• clarify the current execution boundary: the guaranteed output is a dense, copyable local-first handoff artifact, while hosted-builder / Personal Server handshake work remains a later wedge

What's in this PR

• route Create app from Home and import-complete into source overview via URL intent
• add a dedicated source-scoped App Quickstart dialog using the shared Dialog primitive
• add the quickstart state machine with idle, generating, fallback-ready, and error-with-retry states
• add the AppQuickstartArtifact contract plus deterministic fallback artifact generation
• add Copy prompt and Reveal handoff files, writing app-quickstart.md and source-context.json into a dedicated app-data folder
• add the adjacent starter-app slot for the exact-one-live-match case
• add docs for the redesigned quickstart spec, invariants/evals, implementation plan, mermaid flow, and status handoff
• sync the skill-to-evals skill files into this worktree

Execution Boundary

• this PR does not claim that paste-into-Lovable / Bolt / v0 is already a guaranteed happy path
• the current success contract is a truthful, source-aware handoff artifact the user can copy into a local-capable coding environment immediately
• the repo already has real Personal Server grant / tunnel infrastructure for protocol-aware external apps, but packaging that into a reusable hosted-builder quickstart contract is a separate follow-up

Verification

• npx vitest run src/pages/source/index.test.tsx src/pages/source/use-source-overview-page.test.ts src/pages/home/index.test.tsx src/pages/home/components/connected-sources-list.test.tsx
• npm run typecheck
• git diff --check
• cargo check --manifest-path src-tauri/Cargo.toml is still blocked by the existing connector glob / build-script issue noted in the handoff docs

Follow-up

• strengthen the handoff artifact so it is denser, more source-specific, and more explicit about the current local-first execution model
• treat hosted-builder / Personal Server handshake work as a separate wedge rather than smuggling it into this slice