Use regular arena allocation for huge tree nodes.

This avoids grabbing the base mutex, as a step towards fine-grained
locking for huge allocations. The thread cache also provides a tiny
(~3%) improvement for serial huge allocations.

src/jemalloc.c

@@ -240,7 +240,7 @@ a0free(void *ptr)
 	if (chunk != ptr)
 		arena_dalloc(NULL, chunk, ptr, false);
 	else
-		huge_dalloc(ptr);
+		huge_dalloc(NULL, ptr);
 }
 
 /* Create a new arena and insert it into the arenas array at index ind. */