[Feature/Memory] Implement Pre-Allocated eBPF Map Slot Recycler to Suppress Kernel Dynamic Slab Allocations

Commercial Production Objective

Eliminate Linux kernel runtime dynamic slab allocation latency spikes during high-frequency asynchronous socket connection tear-downs under peak multi-tenant AI agent streams .

Bare-Metal Technical Specifications

1. Static Map Allocation Pool: Pre-allocate a high-density, fixed-size ring architecture memory matrix inside the BPF_MAP_TYPE_SOCKMAP stack during the Go proxy daemon initialization phase .
2. Lock-Free Index Reusing: Build an inline atomic slot allocation recycler that maps recycled connection handles immediately without requesting dynamic kernel slab pages.
3. Tail Latency Clamping: Drops dynamic OS kernel memory tracking overhead down to absolute 0%, ensuring pristine line-rate volumetric packet processing boundaries under maximum tenant saturation .

The issue names no files, tests, or concrete entry points. Start by locating the Go proxy daemon initialization and BPF_MAP_TYPE_SOCKMAP setup, then verify whether pre-allocation and lock-free slot recycling are supported by the current design; done means the proposed recycler suppresses the specified dynamic allocations under connection teardown.