Covariance matrix in iterative fitters

[ConnorStoneAstro]

Is your feature request related to a problem? Please describe.
It would be nice to be able to extract the covariance matrix after an iterative fit. Currently it is only possible to update the parameter uncertainties, since only one model is stored in the algorithm at a time.

Describe the solution you'd like
In principle, if it is possible to work with two models at a time then one could iterate through all pairs and build up the full covariance matrix. Similarly for the iterative fit by parameters, one could go through all pairs of chunks and build the covariance matrix.