Change a little bit the benchmarks to be more representative.

bench/bench-stl.cpp

@@ -94,7 +94,7 @@ static void test_rbtree(size_t n)
   for (size_t i = 0; i < n; i++) {
     tree.insert(rand_get());
   }
-    
+  rand_init();
   unsigned int s = 0;
   for (size_t i = 0; i < n; i++) {
     set<unsigned long>::iterator it = tree.find(rand_get());
@@ -115,7 +115,7 @@ test_dict1(unsigned long  n)
   for (size_t i = 0; i < n; i++) {
     dict[rand_get()] = rand_get();
   }
-    
+  rand_init();
   unsigned int s = 0;
   for (size_t i = 0; i < n; i++) {
     map<unsigned long, unsigned long>::iterator it = dict.find(rand_get());
@@ -136,7 +136,7 @@ test_dict2(unsigned long  n)
   for (size_t i = 0; i < n; i++) {
     dict[rand_get()] = rand_get();
   }
-    
+  rand_init();
   unsigned int s = 0;
   for (size_t i = 0; i < n; i++) {
     unordered_map<unsigned long, unsigned long>::iterator it = dict.find(rand_get());
@@ -178,7 +178,7 @@ test_dict_big(unsigned long  n)
     sprintf(s2.a, "%u", rand_get());
     dict[s1] = s2;
   }
-
+  rand_init();
   unsigned int s = 0;
   for (size_t i = 0; i < n; i++) {
     char_array_s s1;