tools: add llama-download for model management
Add a new CLI tool for downloading and managing GGUF models:
- Download from HuggingFace, Docker Hub (OCI), or direct URLs
- Queue multiple downloads with persistent state (download-queue.json)
- Resume interrupted downloads via HTTP Range requests
- Check for model updates via ETag validation (--update)
- Disk space monitoring and validation
- Wait-for-network mode for unreliable connections (--wait-for-network)
- Dry-run mode to preview downloads (--dry-run)
- Support comma-separated repos for batch downloads
Requires LLAMA_CURL=ON to build.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
