Fixing performance benchmark comparison

[cdgriffith]

There is unfortunately a fundamental flaw in the benchmarks, is that they all have the initialization as part of the process. Here is a quick example of how breaking that part out of it changes the benchmarks drastically.

Raw run:

#  <class 'addict.addict.Dict'>  -  get :  1197604715  ops per second.
.#  <class 'addict.addict.Dict'>  -  get through list :  647249139  ops per second.
.#  <class 'addict.addict.Dict'>  -  instantiate :  773292  ops per second.
.#  <class 'addict.addict.Dict'>  -  set :  130434783  ops per second.
.#  <class 'addict.addict.Dict'>  -  set through list :  91240875  ops per second.
.#  <class 'box.box.Box'>  -  get :  57853629  ops per second.
.#  <class 'box.box.Box'>  -  get through list :  25509119  ops per second.
.#  <class 'box.box.Box'>  -  instantiate :  382330  ops per second.
.#  <class 'box.box.Box'>  -  set :  29902815  ops per second.
.#  <class 'box.box.Box'>  -  set through list :  18625442  ops per second.
.#  <class 'scalpl.scalpl.Cut'>  -  get :  71882113  ops per second.
.#  <class 'scalpl.scalpl.Cut'>  -  get through list :  37406484  ops per second.
.#  <class 'scalpl.scalpl.Cut'>  -  instantiate :  367421940  ops per second.
.#  <class 'scalpl.scalpl.Cut'>  -  set :  68681318  ops per second.
.#  <class 'scalpl.scalpl.Cut'>  -  set through list :  40647652  ops per second.
.#  <class 'dict'>  -  get :  1675977835  ops per second.
.#  <class 'dict'>  -  get through list :  746268647  ops per second.
.#  <class 'dict'>  -  instantiate :  612870255  ops per second.
.#  <class 'dict'>  -  set :  1388888605  ops per second.
.#  <class 'dict'>  -  set through list :  798934707  ops per second.