Platform-specific parts library + AI part detection, vendor/image support and CSV import

[TopProjectsCreator]

Motivation

• Separate FTC and Arduino parts workflows so each platform has its own parts library and UI context.
• Prevent repeated AI re-generation of the same part by reusing library entries and caching AI results.
• Make part identification richer by adding vendor page context, image/camera input for material hints, and bulk CSV import that can be AI-enhanced.

Description

• Route open-parts-inventory events from FTCPanel and ArduinoPanel with detail.platform and propagate that through IDELayout to PartsInventoryDialog so the dialog opens in a platform-aware mode.
• Overhaul PartsInventoryDialog to support an activePlatform (FTC/Arduino/General) and show a dedicated header, platform buttons, and platform-scoped filtering.
• Add AI de-duplication and caching in PartsInventoryDialog so identifyWithAI first checks existing inventory and a localStorage AI cache before calling the backend, and cache results per platform.
• Add vendor/product URL input, image upload (file + camera capture), a part preview, and a CSV bulk import flow (parseCsvRows + handleBulkImportCsv) that uses the identify-part function to fill missing metadata and imports many rows in one operation.
• Extend the Supabase function identify-part to accept vendorUrl and imageBase64, fetch basic vendor page metadata when provided, and request a material field in the AI prompt to improve material/part-family inference.

Testing

• Ran lint with npm run lint; linter completed with repository warnings but no new errors introduced.
• Built the app with npm run build; the production build completed successfully.
• Ran unit tests with npm run test; the test suite had 2 failing assertions in src/test/arduinoProductionReadiness.test.ts that appear to be baseline/coverage expectations unrelated to these changes (other tests passed).

---

Codex Task