Provide choice of fitting cost function beyond Chi2

[butler]

Megan Robertson's group from the Universtiy of Houston has indicated the need to use alternate cost functions to the current Chi^{2. This relates mostly to having features in different q ranges that are not always captured if the uncertainties vary widely between q ranges. One should be careful in doing this of course. On the other hand it is clear that if we are ever to have successful co-refinement of SAXS and SANS we will need to do something. Most likely in fact weighting cost function differently for different curves?}