TPOTClassifier.score() doesn't return the same value as TPOTClassifier.fitted_pipeline.score()?

Hello -

I am trying to output the final, fitted pipeline from TPOTClassifier. I am planning on deploying the model to a production environment. From issue #520, it was suggested to pickle the final pipeline:

```
from tpot import TPOTClassifier
#Define the classifier
pipeline_optimizer = TPOTClassifier()
pipeline_optimizer.fit(X, y)
#Output the final pipeline
final_pipeline = pipeline_optimizer.fitted_pipeline_
with open('/tpot_current.pickle', 'wb') as f:
pickle.dump(final_pipeline], f)
```
However, when the output of `pipeline_optimizer.score` and `pipeline_optimize.fitted_pipeline_.score` return different numbers on a new testing data set.

Are the two `.score` functions different, or am I missing something?