llama-sidecar v0.1.0: daemon + benchmarks + eval suite

Go daemon (cmd/llama-sidecar): per-agent llama-server process pool with
LRU eviction, OpenAI-compatible proxy, flag validation (Unsloth port),
deterministic hash-keyed sidecar reuse. Windows service support via
schtasks/NSSM with DETACHED_PROCESS, stdout pipe drain, and request-ctx
decoupled child lifetime.

Bug fixes (3b.1–3b5): -c flag drop from StripShadowingFlags, UTF-8 BOM
in JSON config, -fa → --flash-attn on default, child process exit after
one request (stdin devnull, stdout pipe, CREATE_NO_WINDOW → DETACHED,
context.Background for child lifetime, background reaper goroutine).

bench/: MTP on/off throughput sweep across 8 GGUFs via SSH+schtasks
automation to sam-desktop. Per-GGUF production flags from llama-swap
config with --ctx-size 32768 override.

eval/: accuracy benchmarks (MMLU 100q, GSM8K 50q, HumanEval 164) +
A/B model comparison (14 agent-typed prompts × 8 models). All scripts
resumable at individual question level.

94 Go tests, race detector clean.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

internal/pool/ports_test.go

@@ -0,0 +1,74 @@
+package pool
+
+import (
+	"sync"
+	"testing"
+)
+
+func TestPortAllocator_AllocateRelease(t *testing.T) {
+	pa := NewPortAllocator(8500, 8502)
+	p1, err := pa.Allocate()
+	if err != nil {
+		t.Fatal(err)
+	}
+	p2, err := pa.Allocate()
+	if err != nil {
+		t.Fatal(err)
+	}
+	p3, err := pa.Allocate()
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// All three ports should be distinct
+	if p1 == p2 || p2 == p3 || p1 == p3 {
+		t.Fatalf("expected distinct ports: %d, %d, %d", p1, p2, p3)
+	}
+
+	// Exhausted
+	_, err = pa.Allocate()
+	if err == nil {
+		t.Fatal("expected error when exhausted")
+	}
+
+	// Release and re-allocate
+	pa.Release(p2)
+	p4, err := pa.Allocate()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if p4 != p2 {
+		t.Fatalf("expected released port %d, got %d", p2, p4)
+	}
+}
+
+func TestPortAllocator_Concurrent(t *testing.T) {
+	pa := NewPortAllocator(8500, 8599)
+	var wg sync.WaitGroup
+	allocated := make(chan int, 100)
+
+	for i := 0; i < 100; i++ {
+		wg.Add(1)
+		go func() {
+			defer wg.Done()
+			p, err := pa.Allocate()
+			if err != nil {
+				return
+			}
+			allocated <- p
+		}()
+	}
+	wg.Wait()
+	close(allocated)
+
+	seen := make(map[int]bool)
+	for p := range allocated {
+		if seen[p] {
+			t.Fatalf("duplicate port %d", p)
+		}
+		seen[p] = true
+	}
+	if len(seen) != 100 {
+		t.Fatalf("expected 100 ports, got %d", len(seen))
+	}
+}