Add regularization to map solution

[mpound]

In many cases, the conditions can be ill-posed. Meaning small changes in values or errors can lead to large changes in n and/or G0. This can manifest as stripes in the solution map. The fix for this is to implement some kind of regularization term to the cost function (chisq calculation). Probably the best choice for our case is a regularization term based on the spatial derivatives of the observed ratio maps. This may help force n and G0 to be spatially smooth.