Nate, can you review this for me?

I worry a bit about depending on checking if the child is a heavy footprint as a condition for doing work. It seems like this might be a failure point if the footprints or deblenders are changed. I do not know if there is another good way to do it though.

I agree that it relies on the fact that the current model has children as heavy footprints and parents as footprints. But this is not the only place in our code where we make that assumption.

That is fair enough. Like I said I didn't have a feeling one way or another about that, I just simply wanted to highlight if there was a different way to check then we probably should to future proof things. If not this is fine. Everything looks good to merge to me provided it passes your test, and does not break anything in jenkins. An added test that I think would be good but not required is to run this on some real data (on ci_hsc or your own) just to check against weird failures.

Merged to master.

Bob Armstrong, could you please add a brief (couple of sentences) note to the release notes explaining what this new blendedness metric actually measures? Thanks!

I see this is now in the release notes – thank you!