Fix maxcut computation from samples in qaoa notebook

[danlkv]

Description of Issue (if Applicable)

Fixes #500

What Changes Were Made?

Use reversed key of samples dictionary, to account for littlie-endianness of resulting states

How Was This Tested?

Tested on a graph with following adjacency matrix:

[[0. 1. 0. 1.]
 [1. 0. 1. 1.]
 [0. 1. 0. 1.]
 [1. 1. 1. 0.]]

[CLAassistant]

All committers have signed the CLA.

[danlkv]

Just spotted that my table-of-contents plugin for jupyter notebook added a table of contents, fixed in a second commit

[divshacker]

I suppose this issue was fixed by #1204