Kernel size based on image proportions

[JasonGellis]

It is possible that ripple remover and potentially template matching will benefit from kernel sizes derived from image proportions rather than a 'one size fits all' solution. Currently @crangelsmith has in place an algorithim which seperates profile, dorsal, platform, etc. surfaces based on size. Can we use these to create a function which generates a proportional kernel? A new file - kustom_kernel.py has been created in 'scripts' folder under the scripts branch.