Add M365-backed DOCX to PDF conversion

[bencarver]

Summary

• add a backend conversion service at POST /api/docx-to-pdf that uses Microsoft Graph to upload DOCX to OneDrive, request format=pdf, and return converted PDF bytes
• wire frontend DOCX upload normalization to that endpoint for both Extract and Assembly flows, while keeping PDF behavior unchanged
• improve upload error UX by surfacing detailed conversion failures directly in file rows (missing env, auth/token issues, Graph permission errors)
• add full-stack local dev wiring (dev:backend, dev:full, Vite /api proxy) and document M365 env + permission setup in README.md

Test Plan

• npm run build
• npm run dev:backend starts successfully
• with valid M365 env + granted Graph app permissions, upload a .docx and confirm it appears as converted PDF and can be extracted
• with intentionally broken env, confirm document row displays specific conversion error details

Made with Cursor