Zeropoint and beta

[BrunoSanchez]

Work on fixing the real zeropoint and beta parameters of distribution.

We need a non-linear solver (two implementations in place) for beta, but then fix relative zeropoint scales.
We need if not, an absolute zeropoint calibration between images prior to subtraction.

[BrunoSanchez]

For this it might be useful to improve the cost functions.
Cost functions could leverage source fluxes, by using an aperture or ellipse photometry measurement at each iteration.

[BrunoSanchez]

Additional work could be done with the final flux zeropoint normalization and photometry in Scorr.