refactor(jobs, allocator): optimize queues & dependencies

Major refactor of job system and memory allocator:
- Replaced threadIndex with preferLocal for scheduling
- Switched local queues to SPMCQueue for better performance
- Introduced lock-free JobEdge pool for dependencies
- Removed remainingBatches; use ref counting for completion
- Updated all scheduling APIs and tests to new model
- Optimized FreeList struct sizes and block management
- Added allocation benchmarks
- Disabled OwnershipTransferAnalyzer temporarily
- Bumped assembly versions

Misaki.HighPerformance.Test/Benchmark/AllocationBenchmark.cs

@@ -0,0 +1,58 @@
+using BenchmarkDotNet.Attributes;
+using Misaki.HighPerformance.LowLevel.Buffer;
+using System.Runtime.InteropServices;
+
+namespace Misaki.HighPerformance.Test.Benchmark;
+
+[MemoryDiagnoser]
+public unsafe class AllocationBenchmark
+{
+    private FreeList _frreList;
+
+    private void* _nativePtr;
+    private void* _freeListPtr;
+    private byte[]? _managedArray;
+    private IntPtr _marshalPtr;
+
+    [Params(8, 64, 512, 1024, 4096, 8192)]
+    public nuint Size = 1024;
+
+    [GlobalSetup]
+    public void Setup()
+    {
+        _frreList = new FreeList(8);
+    }
+
+    [GlobalCleanup]
+    public void Cleanup()
+    {
+        _frreList.Dispose();
+    }
+
+    [Benchmark]
+    public void NativeMemoryAlloc()
+    {
+        _nativePtr = NativeMemory.Alloc(Size);
+        NativeMemory.Free(_nativePtr);
+    }
+
+    [Benchmark]
+    public void FreeListAlloc()
+    {
+        _freeListPtr = _frreList.Allocate(Size, 8);
+        _frreList.Free(_freeListPtr);
+    }
+
+    [Benchmark]
+    public void ManagedAlloca()
+    {
+        _managedArray = new byte[Size];
+    }
+
+    [Benchmark]
+    public void GCAlloc()
+    {
+        _marshalPtr = Marshal.AllocHGlobal((nint)Size);
+        Marshal.FreeHGlobal(_marshalPtr);
+    }
+}

Misaki.HighPerformance.Test/Benchmark/ParallelNoiseBenchmark.cs

@@ -45,7 +45,7 @@ public class ParallelNoiseBenchmark
             height = _HEIGHT
         };
 
-        var handle = _jobScheduler.ScheduleParallel(ref job, _LENGTH, 64, -1);
+        var handle = _jobScheduler.ScheduleParallel(ref job, _LENGTH, 64);
         _jobScheduler.Wait(handle);
     }
 

Misaki.HighPerformance.Test/Benchmark/SPMDBenchmark.cs

@@ -36,7 +36,7 @@ public unsafe class SPMDBenchmark
             height = _SIZE,
         };
 
-        var handle = _scheduler.ScheduleParallelSPDM<Jobs.NoiseJobMathSPMD, float>(ref job, _SIZE * _SIZE, 64, -1);
+        var handle = _scheduler.ScheduleParallelSPDM<Jobs.NoiseJobMathSPMD, float>(ref job, _SIZE * _SIZE, 64, false);
         _scheduler.Wait(handle);
     }
 
@@ -50,7 +50,7 @@ public unsafe class SPMDBenchmark
             height = _SIZE,
         };
 
-        var handle = _scheduler.ScheduleParallelFor(ref job, _SIZE * _SIZE, 64, -1);
+        var handle = _scheduler.ScheduleParallelFor(ref job, _SIZE * _SIZE, 64);
         _scheduler.Wait(handle);
     }
 
@@ -64,7 +64,7 @@ public unsafe class SPMDBenchmark
             height = _SIZE,
         };
 
-        var handle = _scheduler.ScheduleParallel(ref job, _SIZE * _SIZE, 64, -1);
+        var handle = _scheduler.ScheduleParallel(ref job, _SIZE * _SIZE, 64);
         _scheduler.Wait(handle);
     }