Migrate voice platform to GCP services

[Usergy-ops]

What changed

• Replaced Azure Blob Storage with private Google Cloud Storage using the VM service account, including authenticated audio streaming and range requests.
• Replaced AWS SES with Resend email delivery.
• Added GCP-ready Cloud SQL, LiveKit, Redis, nginx, TLS renewal, PM2, deployment, and relaunch runbook configuration.
• Removed the committed LiveKit configuration that embedded a secret.

Why

The previous platform depended on expired Azure and unavailable AWS services. This branch moves the production runtime to the already-provisioned GCP project while keeping the application fresh (no database or historic-audio import).

Impact

The platform can be deployed on the GCP VM after Resend and Cloudflare DNS are configured. Audio remains private and is served through signed application links.

Validation

• Focused TypeScript checks for the new GCS storage, Resend email, storage route, and environment configuration.
• Shell syntax validation for deployment, VM setup, and certificate renewal scripts.
• Package manifest validation and staged-diff whitespace check.

A complete production build is still a deployment gate because local Prisma client generation requires its engine download; the GCP deployment script runs generation before building.