Add a non weighted cost function

[Anders Markvardsen]

Suggested by Bill Marshall.

Currently the least squares cost function we have implemented in Mantid is

sum ((y_i^obs-y_ical)/sigma_i)²

which at least in the crystallographic community may be referred to as weighted least squares. In this community another cost function used is

sum ((y_i^obs-y_ical))²

i.e. with all the sigma_i=1

This ticket is to add this latter cost function option to mantid, although should not be the default cost function option in mantid.

The steps for this tickets are

1. Send email to mantid developer list to suggest the name of this cost function to be 'none-weigthed least squares' with the name of the already implemented 'least squares' cost staying the same, partly to ensure backward compatebility... However, someone else may have a different oppinion for good reasons

2. implement this new cost function

[Nick Draper]

Bulk move to assigned at the introduction of the triage step

[Nick Draper]

Moved to R3.5 at the R3.4 code freeze

[Stuart Campbell]

This ticket has been transferred to github issue ​8788