Errors when using reusing relative paths in output section

[desilinguist]

How to reproduce:

1. You want to put logs, models, predictions etc. in the same directory and so you use the same relative path to that directory in the config file.
2. You run the command run_experiment <config_file> from outside the directory where the config file lives.

You will get the error:

Traceback (most recent call last):
  File "/Users/nmadnani/anaconda/envs/garbage/bin/run_experiment", line 6, in <module>
    sys.exit(main())
  File "/Users/nmadnani/anaconda/envs/garbage/lib/python3.4/site-packages/skll/utilities/run_experiment.py", line 108, in main
    ablation=ablation, resume=args.resume)
  File "/Users/nmadnani/anaconda/envs/garbage/lib/python3.4/site-packages/skll/experiments.py", line 697, in run_configuration
    custom_learner_path) = _parse_config_file(config_file)
  File "/Users/nmadnani/anaconda/envs/garbage/lib/python3.4/site-packages/skll/config.py", line 448, in _parse_config_file
    os.makedirs(prediction_dir)
  File "/Users/nmadnani/anaconda/envs/garbage/lib/python3.4/os.py", line 237, in makedirs
    mkdir(name, mode)
FileExistsError: [Errno 17] File exists: '/Users/nmadnani/work/caaspp-wer-garbage-detection/skll/output'

This happens because when checking whether to create the directories in _parse_config_file() in config.py, we check the relative path version instead of the properly located version relative to the config file.

[dan-blanchard]

Linking to #213, since it was a very similar issue.

[desilinguist]

Addressed by #287.