Nate Lust, hate to dump a big review like this on you right now, as I know you tend to attract them, but I have been or will be dumping other big reviews on all of the other candidates.  I'm hoping you'll find some of the modern C++ functional stuff in the pybind11 code more fun than scary, and maybe even have some ideas on how to make it less scary (I think it's actually quite readable and expressive once you've internalized C++ lambda capture syntax, but I'm well aware that's not something most readers of this code will have done).

Almost all changes in afw: https://github.com/lsst/afw/pull/396

with a one-line bugfix in daf_base: https://github.com/lsst/daf_base/pull/42

Note that the afw changes depend on DM-15836, which is also out for review.

Full-stack Jenkins run is now in progress, so there may be small changes for other packages appearing if I discover any breakage.

Quite a few comments, but I trust that you can address them and then merge, as there were no serious algorithmic or logic comments.

Colin Slater, since Nate Lust's review of all of the afw code on this ticket, testing has revealed some (very small) fixes needed in obs_lsstSim and obs_decam.  Could you take a quick look at just those two packages, please?