fix bugs in ipynb2py and app test

[edieYoung]

switch the order of test apps to verify the error.

[edieYoung]

jenkins not passed:
run app tests 2 failed
http://10.239.47.211:18888/view/ZOO-PR/job/ZOO-PR-Python-AppTests/65/console

[edieYoung]

jenkins not passed:
run app tests 4 successfully, test 2 failed:
http://10.239.47.211:18888/view/ZOO-PR/job/ZOO-PR-Python-AppTests/66/console

[edieYoung]

jenkins passed:
test 2 has passed, because the download path was wrong, it worked well after revising to $FTP_URI/analytics-zoo-models-new/object-detection/analytics-zoo_ssd-mobilenet-300x300_PASCAL_0.1.0.model.
test 1 still not passed.
console output:
http://172.168.2.101:8080/view/ZOO-PR/job/ZOO-PR-Python-AppTests/68/console

[edieYoung]

jenkins not passed:
http://172.168.2.101:8080/view/ZOO-PR/job/ZOO-PR-Python-ExampleTests/367/console
textclassification, objectdetection and customized loss and layer have passed.
the predict.py of imageclassification has bug:
File "/opt/work/jenkins/workspace/ZOO-PR-Python-ExampleTests/pyzoo/zoo/examples/imageclassification/predict.py", line 38, in predict
print("\t%s, %f" % (labelMap[sortedProbs[999 - i][1]], sortedProbs[999 - i][0]))
IndexError: list index out of range

[edieYoung]

As for apps test: Revised the download path's filename "variational_encoder" to "variational-encoder", as well as the related calls in the notebook scripts.

[edieYoung]

jenkins passed: app test has finised
http://172.168.2.101:8080/view/ZOO-PR/job/ZOO-PR-Python-AppTests/84/console
time used:
#1 anomaly-detection-nyc-taxi time used:192 seconds
#2 object-detection time used:853 seconds
#3 ncf-explicit-feedback time used:190 seconds
#4 wide_n_deep time used:315 seconds
#5 using_variational_autoencoder_and_deep_feature_loss_to_generate_faces time used:699 seconds
#6 using_variational_autoencoder_to_generate_faces time used:435 seconds
#7 using_variational_autoencoder_to_generate_digital_numbers time used:252 seconds
#8 sentimentAnalysis time used:1487 seconds
#9 image-augmentation time used:131 seconds