prob_classify causing exceptions when working with scikits.

[dchaplinsky]

For some reason underlying method predict_proba from sklearn package is not implemented for some of classifiers (LinearSVC, SGD with loss="hinge") and for some of them it returns result in kind of unexpected format:

Traceback (most recent call last):
  File "classify.py", line 42, in <module>
    neg_prob = classifier.prob_classify(feats).prob("neg")
  File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/nltk/classify/api.py", line 65, in prob_classify
    return self.batch_prob_classify([featureset])[0]
  File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/nltk/classify/scikitlearn.py", line 74, in batch_prob_classify
    return [self._make_probdist(y_proba[i]) for i in xrange(len(y_proba))]
  File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/nltk/classify/scikitlearn.py", line 135, in _make_probdist
    for i, p in enumerate(y_proba)))
TypeError: 'numpy.float64' object is not iterable

Also, to enable predict_proba for SVC/NuSVC you need to pass probability=True to the constructor.