Yaml-based config

[ASvyatkovskiy]

The pull request contains following changes:

• Replace conf.py configuration file which was a member of the main python package with a YAML based config file and a config file parser to allow user making local changes in the configuration without needing to reinstall the package
• Remove dependency on jet_signals
• Move all of the python code from the config to the config parser, which looks like this:

    with open(input_file, 'r') as yaml_file:
        params = yaml.load(yaml_file)
        ....

For instance, the conf['data']['target'] parameter is not an object, but a string. A proper object is created based on the string matching (similarly to the Normalizers):

        if params['target'] == 'hinge':
            params['data']['target'] = t.HingeTarget
        elif params['target'] == 'binary':
            params['data']['target'] = t.BinaryTarget
        elif params['target'] == 'ttd':
            params['data']['target'] = t.TTDTarget
        elif params['target'] == 'ttdlinear':
            params['data']['target'] = t.TTDLinearTarget
        else:
            print('Unkown type of target. Exiting')
            exit(1)

• In addition, restore some of the features that were present in the late version on Julian's github: namely, the positivity mask check and minor changes in the apply_mask for normalization