feat(audit): fan-stress SAT for MSI case-04 fan lag & thermal throttle detection

Two-phase GPU thermal cycling test with per-second telemetry:
- Phases: baseline → load1 → pause (no cooldown) → load2 → cooldown
- Monitors: fan RPM (ipmitool sdr), CPU/server temps (ipmitool/sensors),
  system power (ipmitool dcmi), GPU temp/power/usage/clock/throttle (nvidia-smi)
- Detects throttling via clocks_throttle_reasons.active bitmask
- Measures fan response lag from load start (validates case-04 ~2s lag)
- Exports metrics.csv (wide format, one row/sec) and fan-sensors.csv (long format)
- TUI: adds [F] Fan Stress Test to Health Check screen with Quick/Standard/Express modes

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

audit/internal/tui/forms.go

@@ -1,8 +1,10 @@
 package tui
 
 import (
+	"context"
 	"time"
 
+	"bee/audit/internal/platform"
 	tea "github.com/charmbracelet/bubbletea"
 )
 
@@ -137,6 +139,21 @@ func (m model) updateConfirm(msg tea.KeyMsg) (tea.Model, tea.Cmd) {
 				},
 				pollSATProgress("gpu-amd", since),
 			)
+		case actionRunFanStress:
+			m.busyTitle = "Fan Stress Test"
+			m.progressPrefix = "fan-stress"
+			m.progressSince = time.Now()
+			m.progressLines = nil
+			since := m.progressSince
+			opts := hcFanStressOpts(m.hcMode, m.app)
+			return m, tea.Batch(
+				func() tea.Msg {
+					ctx := context.Background()
+					result, err := m.app.RunFanStressTestResult(ctx, opts)
+					return resultMsg{title: result.Title, body: result.Body, err: err, back: screenHealthCheck}
+				},
+				pollSATProgress("fan-stress", since),
+			)
 		}
 	case "ctrl+c":
 		return m, tea.Quit
@@ -148,9 +165,53 @@ func (m model) confirmCancelTarget() screen {
 	switch m.pendingAction {
 	case actionExportBundle:
 		return screenExportTargets
-	case actionRunAll, actionRunMemorySAT, actionRunStorageSAT, actionRunCPUSAT, actionRunAMDGPUSAT:
+	case actionRunAll, actionRunMemorySAT, actionRunStorageSAT, actionRunCPUSAT, actionRunAMDGPUSAT, actionRunFanStress:
 		return screenHealthCheck
 	default:
 		return screenMain
 	}
 }
+
+// hcFanStressOpts builds FanStressOptions for the selected mode, auto-detecting all GPUs.
+func hcFanStressOpts(hcMode int, application interface {
+	ListNvidiaGPUs() ([]platform.NvidiaGPU, error)
+}) platform.FanStressOptions {
+	// Phase durations per mode: [baseline, load1, pause, load2]
+	type durations struct{ baseline, load1, pause, load2 int }
+	modes := [3]durations{
+		{30, 120, 30, 120},   // Quick:    ~5 min total
+		{60, 300, 60, 300},   // Standard: ~12 min total
+		{60, 600, 120, 600},  // Express:  ~24 min total
+	}
+	if hcMode < 0 || hcMode >= len(modes) {
+		hcMode = 0
+	}
+	d := modes[hcMode]
+
+	// Use all detected NVIDIA GPUs.
+	var indices []int
+	if gpus, err := application.ListNvidiaGPUs(); err == nil {
+		for _, g := range gpus {
+			indices = append(indices, g.Index)
+		}
+	}
+
+	// Use minimum GPU memory size to fit all GPUs.
+	sizeMB := 64
+	if gpus, err := application.ListNvidiaGPUs(); err == nil {
+		for _, g := range gpus {
+			if g.MemoryMB > 0 && (sizeMB == 64 || g.MemoryMB < sizeMB) {
+				sizeMB = g.MemoryMB / 16 // allocate 1/16 of VRAM per GPU
+			}
+		}
+	}
+
+	return platform.FanStressOptions{
+		BaselineSec:  d.baseline,
+		Phase1DurSec: d.load1,
+		PauseSec:     d.pause,
+		Phase2DurSec: d.load2,
+		SizeMB:       sizeMB,
+		GPUIndices:   indices,
+	}
+}