Benchmark 2Q gate decomposition

[jordandsullivan]

The basis_translator (decomposes all gates into a restricted 2 qubit basis set) is the longest pass (43% of runtime), and if we remove it, we save ~40% of the runtime, bringing us to 67% the speed of PyTKET. However, our users will expect 2 qubit gate decomposition in general, and in fact this is the first action item:

1. Compare the speed of this 2 qubit gate decomposition for multi-qubit circuits independently across the different compilers.

Originally posted by @jordandsullivan in #44