Several dataset tests fail instead of skip

[goodfeli]

When you write a test for a dataset, it should skip rather than fail if the user hasn't installed that dataset.
I'm not totally sure how these tests are passing on Travis but tons of them fail on my non-LISA machine:

Traceback (most recent call last):
  File "/usr/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
    self.test(*self.arg)
  File "/usr/local/google/home/goodfellow/pylearn2/pylearn2/datasets/tests/test_wiskott.py", line 12, in     test_wiskott
    data = Wiskott()
  File "/usr/local/google/home/goodfellow/pylearn2/pylearn2/datasets/wiskott.py", line 30, in __init__
    X = 1. - load(path)
  File "/usr/local/google/home/goodfellow/pylearn2/pylearn2/utils/serial.py", line 107, in load
    return np.load(filepath)
  File "/usr/lib/python2.7/dist-packages/numpy/lib/npyio.py", line 370, in load
    fid = open(file, "rb")
IOError: [Errno 2] No such file or directory:        '/usr/local/google/home/goodfellow/data/wiskott/wiskott_fish_layer0_15_standard_64x64_shuffled.npy'

[goodfeli]

Looks like 41 tests fail in this way for me right now

[lamblin]

The tests are skipped if PYLEARN2_DATA_PATH is not defined, that is how they are passing on Travis.
Indeed, the individual tests should also be skipped if the corresponding dataset is not installed.