feat: redesign ai-chat-memory-template using native flowsh capabilities

✨ COMPLETE REDESIGN to showcase flowsh's strengths:
- Replace complex jq/shell operations with native flowsh nodes
- Use variable-assignment, variable-aggregation, llm, and code nodes effectively
- Maintain core functionality: memory persistence, context awareness, conversation tracking
- Eliminate shell syntax issues through proper node architecture

🏗️ ARCHITECTURE IMPROVEMENTS:
- 11 nodes, 10 edges (clean linear flow)
- Native flowsh patterns instead of circumventing with complex code
- Proper LLM integration with conversation context
- Simple file-based persistence without complex JSON manipulation

🎯 DEMONSTRATES FLOWSH CAPABILITIES:
- Variable management and templating
- LLM integration with context
- Data aggregation and flow
- Persistent state management

🎉 ACHIEVEMENT: 40/40 templates now pass shell syntax validation!

Author: Tom Brandenburg