Make the shiftpush op a bunch more efficient when handling more than …

…a single element. We still call it with 1 far, far too often, but this helps for the other cases. for 1..100000 { } gets 3 times faster with this patch, for example.

src/ops/perl6.ops

@@ -1064,21 +1064,18 @@ simply discarded.
 inline op perl6_shiftpush(inout PMC, in PMC, in INT) :base_core {
     INTVAL count = $3;
     INTVAL elems = VTABLE_elements(interp, $2);
-
     if (count > elems) count = elems;
 
     if (!PMC_IS_NULL($1)) {
+        INTVAL pos = 0;
         while (count > 0) {
-            VTABLE_push_pmc(interp, $1, VTABLE_shift_pmc(interp, $2));
-            count--;
-        }
-    }
-    else {
-        while (count > 0) {
-            VTABLE_shift_pmc(interp, $2);
+            VTABLE_push_pmc(interp, $1, VTABLE_get_pmc_keyed_int(interp, $2, pos));
             count--;
+            pos++;
         }
     }
+    if ($3 > 0)
+        VTABLE_splice(interp, $2, pmc_new(interp, enum_class_ResizablePMCArray), 0, $3);
 }