MLX native restored-cache primitive — systemic fix for the Mac throughput collapse#110

FluffyAIcode wants to merge 4 commits into

AgentMemory/v04-mlx-port-incremental-decode-2815FluffyAIcode/Kakeya-LLM-Inference-engine:AgentMemory/v04-mlx-port-incremental-decode-2815from

AgentMemory/v04-mlx-native-cache-primitive-2815FluffyAIcode/Kakeya-LLM-Inference-engine:AgentMemory/v04-mlx-native-cache-primitive-2815Copy head branch name to clipboard