Generation of `Experiment` covariance matrix

Created by: nhartland

So the first step that we have agreed in the whole theory errors business is the separation of covariance matrix generation logic from the Experiment class.

I can think of two ways to proceed:

• Have a function GenerateCovMat(vector<CommonData>, vector<ThPredictions>) which generates a joint covariance matrix for a set of commondata (and their corresponding t0 predictions). The logic here would be more or less the same as in the current Experiment function.

• Have a function that can concatenate CommonData i.e a new constructor CommonData::CommonData(vector<CommonData>) and use the existing GenerateCovMat(CommonData, ThPredictions).

I have a slight preference for the second option, simply as then everything onwards (computing chi2-values etc) becomes trivial as you have a 1-to-1 correspondence between a single CommonData object and a single covariance matrix.

I'd like to implement this pretty soon, as @rabah-khalek could do with it. Does anyone have any strong opinions?