Fixed the bottleneck in profiling.

profiling/test.jl

 using BenchmarkTools
 using Random
+using Profile
 
 function f()
+
   lst = rand(50000)
   for i in lst
     if i > 0.99
@@ -9,6 +11,7 @@ function f()
       print(s)
     end
   end
+
 end
 
 f()
@@ -16,3 +19,7 @@ f()
 x = @benchmark f()
 println()
 println(x)
+
+@profile f()
+Profile.print()
+Profile.clear()