pvalue/gof, fwhm correlated uncertainty, iterative fit, SPD covmat, test data, norm. residuals

[LisaSchlueter]

• gof: pvalue based on beast-squares (baseline), likelihood ratio, MC
• fwhm with correlated uncertainties
• iterative fit in case cov mat not semi positive definite
• function to find nearest semi positive definite cov mat
• MC test data (energy spectrum based on randomized model function)
• normalized residuals

[theHenks]

@LisaSchlueter I left some comments and review proposals. If you could resolve these, we can merge. Most of them just minor things.