Updated lnprob function with logf

[jontymarshall]

I implemented the nuiscance parameter "logf" in the lnprob function within measure_extinction/utils/fit_model.py following the emcee docs here (https://emcee.readthedocs.io/en/stable/tutorials/line/). This enables the fitting to take some steps toward addressing if the error bars are underestimated. To do this I converted the existing 'self.weights' back into a numpy array of errors before summing them with the model and logf as required.

I also added "logf" to the params list as the final element to assist in propagating the new parameter through the code.

[karllark]

Thanks for this PR. Need to think/evaluate what adding this would do overall.