proxy,ui: add performance monitoring with Prometheus metrics (#743)

Add a comprehensive performance monitoring system that collects CPU, memory, swap, load average, network IO, and GPU stats. Provides both a REST API for the UI and a Prometheus /metrics endpoint.

Backend changes:
- New internal/perf package with configurable interval-based stats collection
- GPU monitoring via LACT (Unix socket) and nvidia-smi fallback on Linux
- Ring buffer (internal/ring) for time-series stat storage
- Prometheus /metrics endpoint with all system and GPU metrics
- Moved LogMonitor to internal/logmon package
- New PerformanceConfig for hot-reloadable monitoring settings
- REST /api/performance endpoint replacing SSE streaming

UI changes:
- New Performance page with real-time charts for CPU, memory, GPU, and network
- Reusable PerformanceChart component
- LLAMA_SWAP_URL environment variable support
- Improved capture dialog display

Other:
- Example Grafana dashboard for Prometheus metrics
- monitor-test standalone binary
- Config schema and example updates

fixes #596

internal/ring/buffer.go

@@ -0,0 +1,39 @@
+package ring
+
+type Buffer[T any] struct {
+	buf  []T
+	head int
+	size int
+}
+
+func NewBuffer[T any](capacity int) Buffer[T] {
+	if capacity < 1 {
+		capacity = 1
+	}
+	return Buffer[T]{buf: make([]T, capacity)}
+}
+
+// Push adds v, overwriting the oldest entry when the buffer is full.
+func (r *Buffer[T]) Push(v T) {
+	cap := len(r.buf)
+	if r.size < cap {
+		r.buf[(r.head+r.size)%cap] = v
+		r.size++
+	} else {
+		r.buf[r.head] = v
+		r.head = (r.head + 1) % cap
+	}
+}
+
+// Slice returns all entries in insertion order as a new slice.
+func (r *Buffer[T]) Slice() []T {
+	if r.size == 0 {
+		return nil
+	}
+	cap := len(r.buf)
+	result := make([]T, r.size)
+	for i := 0; i < r.size; i++ {
+		result[i] = r.buf[(r.head+i)%cap]
+	}
+	return result
+}

internal/ring/buffer_bench_test.go

@@ -0,0 +1,44 @@
+package ring
+
+import "testing"
+
+const benchCap = 600 // matches default MaxAge/Every (1min / 100ms)
+
+func BenchmarkBuffer_PushNoWrap(b *testing.B) {
+	for b.Loop() {
+		buf := NewBuffer[int](b.N + 1)
+		for i := range b.N {
+			buf.Push(i)
+		}
+	}
+}
+
+func BenchmarkBuffer_PushWrap(b *testing.B) {
+	buf := NewBuffer[int](benchCap)
+	b.ResetTimer()
+	for i := range b.N {
+		buf.Push(i)
+	}
+}
+
+func BenchmarkBuffer_Slice(b *testing.B) {
+	buf := NewBuffer[int](benchCap)
+	for i := range benchCap {
+		buf.Push(i)
+	}
+	b.ResetTimer()
+	for range b.N {
+		_ = buf.Slice()
+	}
+}
+
+func BenchmarkBuffer_PushAndSlice(b *testing.B) {
+	buf := NewBuffer[int](benchCap)
+	b.ResetTimer()
+	for i := range b.N {
+		buf.Push(i)
+		if i%benchCap == 0 {
+			_ = buf.Slice()
+		}
+	}
+}

internal/ring/buffer_test.go

@@ -0,0 +1,65 @@
+package ring
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestBuffer_EmptySliceIsNil(t *testing.T) {
+	b := NewBuffer[int](4)
+	assert.Nil(t, b.Slice())
+}
+
+func TestBuffer_PushBelowCapacity(t *testing.T) {
+	b := NewBuffer[int](4)
+	b.Push(1)
+	b.Push(2)
+	assert.Equal(t, []int{1, 2}, b.Slice())
+}
+
+func TestBuffer_PushAtCapacity(t *testing.T) {
+	b := NewBuffer[int](3)
+	b.Push(1)
+	b.Push(2)
+	b.Push(3)
+	assert.Equal(t, []int{1, 2, 3}, b.Slice())
+}
+
+func TestBuffer_PushOverCapacityEvictsOldest(t *testing.T) {
+	b := NewBuffer[int](3)
+	b.Push(1)
+	b.Push(2)
+	b.Push(3)
+	b.Push(4)
+	assert.Equal(t, []int{2, 3, 4}, b.Slice())
+}
+
+func TestBuffer_CapacityOne(t *testing.T) {
+	b := NewBuffer[int](1)
+	b.Push(1)
+	b.Push(2)
+	assert.Equal(t, []int{2}, b.Slice())
+}
+
+func TestBuffer_ZeroCapacityDefaultsToOne(t *testing.T) {
+	b := NewBuffer[int](0)
+	b.Push(42)
+	assert.Equal(t, []int{42}, b.Slice())
+}
+
+func TestBuffer_SliceReturnsCopy(t *testing.T) {
+	b := NewBuffer[int](4)
+	b.Push(10)
+	s := b.Slice()
+	s[0] = 99
+	assert.Equal(t, []int{10}, b.Slice())
+}
+
+func TestBuffer_InsertionOrderPreservedAfterWrap(t *testing.T) {
+	b := NewBuffer[int](4)
+	for i := 1; i <= 8; i++ {
+		b.Push(i)
+	}
+	assert.Equal(t, []int{5, 6, 7, 8}, b.Slice())
+}