Enhanced local worker integration (v2.1.0)

[typelicious]

Goal

Complete local worker integration with Grid auto-discovery, enhanced health metrics, and dynamic model enumeration for v2.1.0.

Requirements

• Grid integration for automatic worker discovery across network
• Enhanced health metrics: GPU utilization, memory usage, queue depth monitoring
• Dynamic model enumeration: Query endpoints of local workers to auto-discover available models
• Lifecycle management hooks: Start/stop/restart capabilities for local workers
• Integration with routing core: Use local worker metrics for intelligent routing decisions

Context

From v2.1.0 planning in FAIGATE-ROADMAP.md:

• Local worker completion is a core theme
• Builds on basic local worker support added in v2.0.0
• Should provide operators with full visibility and control over local resources

Tasks

• Grid integration for network-wide worker discovery
• Enhanced health metrics collection (GPU, memory, queue)
• Dynamic model enumeration from endpoints
• Lifecycle management API for local workers
• Dashboard integration for local worker monitoring
• Routing integration using local worker metrics
• Documentation and operational guides

Notes

• Complements existing local worker examples (Ollama, vLLM, LM Studio, LiteLLM)
• Enables true hybrid cloud/local routing intelligence
• Maintains gateway-first architecture while enhancing local capabilities