From Cascade2d pooling/stride traits don't make sense, remove them

[blasscoc]

#: Subsequent convolutions can be applied to downsampled images for
#  efficiency.
# Provide some options with Keras for this:
# Max - MaxPooling (take max value in a window)
# Average - AveragePooling (average values in window)
pooling_type = Enum(["none", "max", "average"])

# Stride - Set a stride when applying the convolutions:
# Interpreted as "factor 2^n", to apply at each convolution.
# each step, "0" gives a stride of 1 sample (the default).
# 1 will apply convolutions at every second sample.
# For now, negative numbers are not valid.
stride_log2 = Int(0)