More TODOs added

EvenPerfectBitScanner/Program.cs

@@ -940,6 +940,7 @@ internal static class Program
 							continue;
 						}
 
+						// Can we prepare as many buffers as many threads were requested upfront and have them prepared in separate task, so that they're always ready when needed?
 						activeCount = BuildPrimeBuffer(divisor, primeValues, allowedMaxValues, stateFlags, primeBuffer, indexBuffer, completionsBuffer, ref completionsCount, ref remainingStates, activeStateMask, ref activeStartIndex);
 
 						if (completionsCount > 0)

PerfectNumbers.Core/Gpu/MersenneNumberDivisorByDivisorGpuTester.cs

@@ -478,6 +478,7 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester
 					accelerator.Synchronize();
 					resultView.CopyToCPU(ref MemoryMarshal.GetReference(exponentSlice), computeCount);
 
+					// TODO: Can we move this somehow to GPU? MulMod64 takes half of the time in CheckDivisor
 					for (int i = 0; i < computeCount; i++)
 					{
 						int position = positionSpan[i];