Fix 'transposition growth' computation for older compilers

The issue is probably due to some shenanigans with float-versus-double or converting size_t to
double including some rounding up. Either way, it may introduce undefined behaviour or be
interpreted multiple ways. This leads GCC 10 to unexpectedly abort repeated transposition.

src/adiar/internal/algorithms/quantify.h

@@ -908,11 +908,11 @@ namespace adiar::internal
       const size_t dd_size = dd.size();
 
       // Do Partial Quantification as long as...
-      //   1. ... it stays smaller than 1+epsilon% of the input size.
-      const size_t transposition__size_threshold = static_cast<size_t>(
-        std::min<double>(std::numeric_limits<size_t>::max(),
-                         ep.template get<exec_policy::quantify::transposition_growth>()
-                           * static_cast<double>(dd_size)));
+      //   1. ... it stays smaller than 1+epsilon of the input size.
+      const size_t transposition__size_threshold =
+        (std::min(static_cast<double>(std::numeric_limits<size_t>::max() / 2u),
+                  static_cast<double>(ep.template get<exec_policy::quantify::transposition_growth>())
+                  * static_cast<double>(dd_size)));
 
       //   2. ... it has not run more than the maximum number of iterations.
       const size_t transposition__max_iterations =