[SYCL][Graph] Refine E2E test comment (#16842)

The code comment in
`Update/FreeFunctionKernels/update_with_indices_ordering.cpp` mentions
that it is unknown why the execution time is slower than the equivalent
`Update/update_with_indices_ordering.cpp` test without free functions.

This has now been investigated in CMPLRLLVM-64841 and has found to be
due to the explicit arguments in the free function signature avoiding
constant folding from lambdas. Note that we want to keep these
parameters as the test is stressing the case where multiple args are
passed to the free function.

Author: EwanC

sycl/test-e2e/Graph/Update/FreeFunctionKernels/update_with_indices_ordering.cpp

@@ -21,8 +21,9 @@ int main() {
   // Use a large N to try and make the kernel slow
   const size_t N = 1 << 16;
 
-  // Reduce amount of work compared to version of test without free functions
-  // due to CMPLRLLVM-64841
+  // Reduce amount of work compared to version of test with lambdas,
+  // as using explicit parameters in the free function signature results
+  // in slower IR than is created from constant folding in the lambda.
   const size_t NumKernelLoops = 1;
   const size_t NumSubmitLoops = 1;