new bench that tests random command ordering, to confuse the branch predictor

[fabi321]

The current implementation takes a significant hit from a random load, as it turns out that the branch predictor is pretty good at predicting predictable stuff.

(11ms vs 19ms on my machine)

[sbernauer]

Related: #4
Thanks a lot!

[sbernauer]

Just for the record:

parse_draw_commands_ordered/1920 x 1080
                        time:   [25.405 ms 25.929 ms 26.617 ms]
Found 3 outliers among 100 measurements (3.00%)
  3 (3.00%) high severe

parse_draw_commands_shuffled/1920 x 1080
                        time:   [32.665 ms 32.684 ms 32.705 ms]
Found 4 outliers among 100 measurements (4.00%)
  1 (1.00%) high mild
  3 (3.00%) high severe