docs: add orchestration versioning best practices guide

[affandar]

Summary

Adds a practical guide for managing versioned orchestrations in production.

What's Included

• Code Organization Pattern - Keeping NAME constants stable, version suffix functions, trace logging with version prefix
• Registration Pattern - register_typed() vs register_versioned_typed() examples
• Version Upgrade Timing (Critical) - Explains that upgrades happen at continue_as_new time, not server restart, with timeline example
• Best Practices Checklist - DOs and DON'Ts for versioning
• Version Info Flow - How version info propagates from Duroxide → Management API → UI
• Common Scenarios - Hot-fixes, state migration, version pinning

Why This Is Needed

The existing migration-guide.md covers breaking changes between Duroxide versions, but doesn't cover practical patterns for versioning your own orchestrations in production. This came from real-world experience implementing long-running orchestrations that needed version upgrades without disrupting in-flight instances.

---

Note

Introduces versioning guidance for long-lived orchestrations and how upgrades occur safely.

• Adds Orchestration Versioning section to ORCHESTRATION-GUIDE.md covering registration (register_typed vs register_versioned_typed), upgrade timing at continue_as_new, and concise best practices
• Creates production-patterns.md with detailed patterns: code organization by stable NAME + version-suffixed functions, registry setup, upgrade timing timeline, DO/DON'T checklist, version info flow (provider → client → API → UI), and practical scenarios (hot-fix, state migration, version pinning)
• Updates table of contents in ORCHESTRATION-GUIDE.md to include the new sections and link to production-patterns.md

^{Written by Cursor Bugbot for commit 09ff705. This will update automatically on new commits. Configure here.}