Current image representation uses more space than neccesary

[Brian-ED]

The current image representation doesn't account for the fact that CBQN doesn't store arrays as unsigned int, it stores them as signed, so the current representations use 2x more space than anticipated.
but also, it isn't perfect anyways since if you have a format that conserves 1 bit for alpha and 8 for red then the alpha will still use 8 bits.

A simple solution that i will implement is to just always use binary arrays and have a converter to floats and back for when you want to operate on the image representations...

I don't like how this overcomplicates it though, i wish there was a way to store RGBA seperated neatly