GaussianCentroid is a measurement algorithm in meas_base that fits a 2D Gaussian to sources to determine the centroid. We have recently discovered that the code has bugs (DM-13299).

I propose dropping GaussianCentroid, which will save on maintenance effort. Since SdssCentroid performs at least as well, is faster, and (as far as I am aware) is always preferred by developers and users, I believe the only reason not to drop GaussianCentroid is to keep a second centroiding algorithm in the main stack. I suggest that if we find that we need a second centroiding algorithm in the future, the code could be resurrected from git.