feat: add ISlideComputer interface and GPU implementation

ISlideComputer abstracts batched directional-distance computation so GPU
implementations can process all slide offsets in a single kernel launch.
GpuSlideComputer uses ILGPU with prepared edge data (precomputed inverse
deltas and min/max bounds) and caches stationary/moving buffers across
calls. GpuEvaluatorFactory exposes a singleton factory method.

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

OpenNest.Gpu/GpuEvaluatorFactory.cs

@@ -11,6 +11,8 @@ namespace OpenNest.Gpu
         private static bool _probed;
         private static bool _gpuAvailable;
         private static string _deviceName;
+        private static GpuSlideComputer _slideComputer;
+        private static readonly object _slideLock = new object();
 
         public static bool GpuAvailable
         {
@@ -46,6 +48,29 @@ namespace OpenNest.Gpu
             }
         }
 
+        public static ISlideComputer CreateSlideComputer()
+        {
+            if (!GpuAvailable)
+                return null;
+
+            lock (_slideLock)
+            {
+                if (_slideComputer != null)
+                    return _slideComputer;
+
+                try
+                {
+                    _slideComputer = new GpuSlideComputer();
+                    return _slideComputer;
+                }
+                catch (Exception ex)
+                {
+                    Debug.WriteLine($"[GpuEvaluatorFactory] GPU slide computer failed: {ex.Message}");
+                    return null;
+                }
+            }
+        }
+
         private static void Probe()
         {
             _probed = true;