Initial PollardRho implementation for GPU. Added pool for accelerator streams shared across the threads. GPU processing performance and stabilization improvements. GpuUInt128.BinaryGcd switched to the version which outperformed other in benchmarks.

PerfectNumbers.Core/PerfectNumberConstants.cs

@@ -12,7 +12,8 @@ public static class PerfectNumberConstants
     public static readonly uint PrimesLimit = 1_000_000; //(ulong)Array.MaxLength;// 1_000_000;
     public const int MaxOddPowersCount = 128;
 	public const int PooledArrayThreshold = 64;
-	public static readonly int RollingAccelerators = 128; //SharedGpuContext.Device.MaxNumThreadsPerGroup;
+	public static readonly int RollingAccelerators = 256; //SharedGpuContext.Device.MaxNumThreadsPerGroup;
+	public const int ThreadsByStream = 1;
     // TODO: Load these limits from the benchmark-driven configuration so CPU and GPU scans stay aligned with the optimal
     // divisor-cycle datasets we generate offline.
     // TODO: Promote these magic numbers into a runtime profile derived from EvenPerfectBitScanner.Benchmarks so we can retune