app.js needs to be restructured and new functions built

[BryanSouza91]

app.js should focus around the TensorFlow model prediction functionality. As such I think we should separate the parts of the process into various functions, so that common ones can be easily reused.

Decode function is reusable, this might be the entire pre-processing function, also still need to work on the Channel Norm.
Predict function is reusable, this will be used by the test images provided or by the user input images.
renderImageLabel function is reusable, this displays the predicted label along with the accuracy.

will need to add in the Webcam functionality later.

[BryanSouza91]

some links:
use this to get imageData to use with TF from canvas
THIS fileUploadBlob to ImageData
a way to call the predict function from the file input form

[BryanSouza91]

should close this issue, Do you agree, Graham?