Run first experiments and analyze first results

[kudkudak]

Sanity checks:

• We are not picking extreme hyperparameter values (model hyperparameters)
• Is computationally ok (do we have have any computational bottleneck?)
• Check if memory usage is a problem
• Add plotting functions
• Curves of different strategies meet in single point
• Repeatitive
• All strategies (including CSJ)

[igorsieradzki]

• Picking extreme parametrs checked, none were picked
• Computationally it looks ok, most of the iteration time is used for fitting grid search (see below)
• No memory issues as of now
• Added plotting functions for curves and times charts
• Checked that same experiment params yield same results
• All strategies (except new UCT) are implemented

Curves (disclaimer: QGB is not fitted, I've just picked some random C, hence the poor performance)

Times for random experiment per strategy:

[igorsieradzki]

Valid and Aleph results: https://gist.github.com/igorsieradzki/4505020233b18aae971b