So I went to use the gaussian with rayon and noticed it was horribly busted.

Whilst I was in there:

• I opted for the gAussian... spelling, I am pretty sure that's not just an Aussie thing because wikipedia spells it that way too.
• a safe_get() -> Option<Dn>
• a helper to get a ref to the ImageBuffers in an Image, which for some reason we didn't already have.

Image::buffers(&self) -> &[ImageBuffer] 

• I added a benchmark for the st vs the rayon version of gaussian_blur, on my 2013 macbook the results are:

GaussianBlur/single-threaded time: 1.0053 s
GaussianBlur/rayon  time:  303.55 ms 

And on my bigger rig (AMD Ryzen 5950x) its' more like ~380ms for the ST and ~160ms for the Rayon version. [on 100 samples]

To run the benchmarks on your own machine(s) do a cargo bench --features='rayon', there are graphs in the ./target/criterion/* directory that gets created by this bench harness .

I was working on this because I am doing some GPGPU image processing at the moment and prefer to play with space data when playing, but that's another PR for another time.