Better explanation on testing

Hi, first of all, thanks for the great work! It was really easy to use.
I had a problem with refining my images with a pretrained model though.

In readme you say: "To refine all synthetic images with a pretrained model":

$ python main.py --is_train=False --synthetic_image_dir="./data/gaze/UnityEyes/"

You are missing the `load_path` argument. Apparently, it is path to the __directory__ with models, and it is __relative to `logs` dir__. For example:

$ python main.py --is_train=False --synthetic_image_dir="./data/gaze/UnityEyes/" --load_path generative_2017-03-07_01-40-07

There is no indication whatsoever, that the model is loaded rather than initialized _during testing_. And of course if it is initialized then it will simply write garbage to refined images, leaving you wondering.

I don't know though how to make it clear whether the model was loaded or initialized if you use `tf.train.Supervisor`, maybe specify `wait_for_checkpoint=True` when calling `prepare_or_wait_for_session`.

Start with the README section titled "To refine all synthetic images with a pretrained model" and compare its command with the reported main.py invocation using load_path. Clarify the model-directory path and whether testing loads or initializes a model; done means the documented command and behavior no longer leave pretrained-model users uncertain.

Written by the indexing model from the issue text.