Badness space update

I've been talking to some maths teachers at the institute where I work. They've proved the identity I needed to show that the determinant form of the TOP-RMS error is correct. It'll take some time to write up but I'm satisfied with it.

That means badness is proportional to the matrix volume of a space of weighted mappings with their means subtracted. Alternatively, a geometry that measures lengths as standard deviations and cosines as correlation coefficients.

Graham