Improve local dev startup resilience across backend, web, and mobile

[Arun-kushwaha007]

Summary

This PR improves local development startup behavior across the stack.

It fixes the backend Redis failure loop, makes the web app handle backend outages more cleanly, and aligns
the mobile app with the current Expo package expectations and typings.

Changes

Backend

• Prevented BullMQ queue producers from eagerly failing when Redis is unavailable
• Disabled BullMQ workers gracefully when Redis cannot be reached
• Disabled notification pub/sub cleanly when Redis is unavailable
• Made Redis cache reads/writes fall back safely instead of throwing during local startup

Web

• Returned 503 responses from auth proxy routes when the backend is unavailable instead of throwing
  500
• Made auth token refresh middleware fail safely when backend requests cannot complete
• Added missing /privacy and /terms pages to remove linked-page 404s during dev

Mobile

• Updated Expo-related package versions to match the installed Expo SDK expectations
• Fixed stricter typing issues introduced by those package updates
• Updated notification handler typing for the newer Expo notifications API
• Fixed UI component typing around gradients and style props

Verification

Ran:

• pnpm --filter backend lint
• pnpm --filter backend test
• pnpm --filter web lint
• pnpm --filter mobile lint

Commits

• fix(backend): tolerate missing Redis during local startup
• fix(web): handle backend outages and add legal pages
• chore(mobile): align Expo packages and strict typings

Notes

• Backend now gets past the Redis connection-refused loop and continues to Prisma startup.
• The remaining backend startup dependency is the local PostgreSQL instance configured at
  localhost:5432.
• Expo dev startup still depends on network access in this environment because Expo CLI fetches remote
  metadata during startup.