Add distinction between additive and multiplicative errors

[shunt16]

From Zhav Loizeau in #1:

Also, should there be a “how” uncertainty variable is linked to observation variable? I know we mostly deal with additive errors but should there be some way to say if some error is multiplicative for example?

[shunt16]

Any thoughts @pdevis?

[EmmaWoolliams]

"Multiplicative errors" are generally expressed through relative uncertainties, and "additive errors" through absolute uncertainties (in same units). In a statistical model the former would be (1+D)*x and the latter as x+D. So I think this is covered by the relative / absolute stuff