Feature Request: Other padding modes for convolve

[awillats]

Currently only the 'valid' mode of convolution is supported. I'd like to help add support for the 'same' and 'full' padding modes included in most other implementations of convolution:

visual explanation of convolution padding modes
matlab implementation
numpy implementation
a little more discussion

I've hacked together something simple for the problem I have that implements this, (and verified it against numpy's implementation) so I'm happy to help with a pull request! But I'd need a little help formatting it in a clean consistent way (not sure how to set up the spec.js etc.)