Inline GPU capability checks for pow2 heuristics

PerfectNumbers.Core/Gpu/PrimeOrderGpuHeuristics.cs

@@ -73,7 +73,7 @@ internal static class PrimeOrderGpuHeuristics
 
         PrimeOrderGpuCapability capability = s_capability;
 
-        if (!SupportsPrime(prime, capability))
+        if (prime.GetBitLength() > capability.ModulusBits)
         {
             overflowRegistry[prime] = 0;
             return GpuPow2ModStatus.Overflow;
@@ -81,7 +81,7 @@ internal static class PrimeOrderGpuHeuristics
 
         for (int i = 0; i < exponents.Length; i++)
         {
-            if (!SupportsExponent(exponents[i], capability))
+            if (exponents[i].GetBitLength() > capability.ExponentBits)
             {
                 return GpuPow2ModStatus.Overflow;
             }
@@ -91,16 +91,6 @@ internal static class PrimeOrderGpuHeuristics
         return computed ? GpuPow2ModStatus.Success : GpuPow2ModStatus.Unavailable;
     }
 
-    private static bool SupportsPrime(ulong prime, PrimeOrderGpuCapability capability)
-    {
-        return prime.GetBitLength() <= capability.ModulusBits;
-    }
-
-    private static bool SupportsExponent(ulong exponent, PrimeOrderGpuCapability capability)
-    {
-        return exponent.GetBitLength() <= capability.ExponentBits;
-    }
-
     private static bool TryComputeOnGpu(ReadOnlySpan<ulong> exponents, ulong prime, Span<ulong> results)
     {
         try

todo.txt

@@ -1,3 +1,2 @@
-- Inline the PrimeOrderGpuHeuristics capability checks at each call site and delete the temporary SupportsPrime/SupportsExponent helpers once the overflow gating is stable.
 - Swap PrimeOrderGpuHeuristics.Pow2ModCpu for the existing Pow2MontgomeryModWindowed helper and plan the same windowed ladder for the GPU kernel implementation.
 - Extend the GPU numeric type probe to cover proposed custom types such as GpuUInt256 and GpuUInt512 once their definitions are available so we can plan the wide-order heuristics on the device.