feat(tui): live GPU chart during stress test, full VRAM allocation

- GPU Platform Stress Test now shows a live in-TUI chart instead of nvtop.
  nvidia-smi is polled every second; up to 60 data points per GPU kept.
  All three metrics (Usage %, Temp °C, Power W) drawn on a single plot,
  each normalised to its own range and rendered in a different colour.
- Memory allocation changed from MemoryMB/16 to MemoryMB-512 (full VRAM
  minus 512 MB driver overhead) so bee-gpu-stress actually stresses memory.

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

audit/internal/tui/forms.go

@@ -182,12 +182,13 @@ func hcFanStressOpts(hcMode int, application interface {
 		}
 	}
 
-	// Use minimum GPU memory size to fit all GPUs.
+	// Use nearly full GPU memory on the smallest GPU (leave 512 MB for driver overhead).
 	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
+			free := g.MemoryMB - 512
+			if free > 0 && (sizeMB == 64 || free < sizeMB) {
+				sizeMB = free
 			}
 		}
 	}