Test GP-based optimization

[arogozhnikov]

Build an additional test that GaussianProcess optimization works fine.

For batch optimization, probably some improvement is possible, which also minimizes an overall variance as well as looking for most optimal model. (Need to check if some well-tested implementation exists).

Sklearn's gaussian processes don't support variance of measurements, which probably could improve search.

[arogozhnikov]

related link:
http://arxiv.org/pdf/1206.2944v2.pdf

[arogozhnikov]

Currently simple GP-optimization is demonstrated in https://github.com/yandex/rep/blob/master/howto/03-howto-gridsearch.ipynb via RegressionOptimization, but maybe some specific solutions will be added later.