Bugfix/cv folds file logging

[Lguyogiro]

• Addresses issue Weird logging issue when specifying cv_folds_file #349 . Previously, cv_folds was assumed to be an int. But if "cv_folds_file" is specified in the config, cv_folds will be a dictionary mapping ids to fold numbers and requires new code to correctly log the number of folds.
• Fixed by checking the type of cv_folds and, if it is not an int, logging the number of unique fold ids from the dict values.
• Added a cv folds file to tests/train/ and test_cv_folds_file_logging to test suite

[coveralls]

Coverage decreased (-0.03%) to 91.251% when pulling 9e308a9 on bugfix/cv_folds_file-logging into ac4df66 on master.

[coveralls]

Coverage increased (+0.6%) to 91.865% when pulling d17f328 on bugfix/cv_folds_file-logging into ac4df66 on master.

[desilinguist]

Looks good. Thanks @Lguyogiro!

[Lguyogiro]

I've pushed one more minor change: previously the experiment name (in the config file) for testing the cv folds file logging was "test_save_cv_folds". I've updated to reflect what it is actually testing.

[coveralls]

Coverage increased (+0.5%) to 91.788% when pulling c816bdc on bugfix/cv_folds_file-logging into ac4df66 on master.

[coveralls]

Coverage increased (+0.6%) to 91.865% when pulling c816bdc on bugfix/cv_folds_file-logging into ac4df66 on master.

[desilinguist]

Since this is a very simple change, I am just going to merge it now without waiting for another approval.