The code is complete (within agreed scope) and a technote (dmtn-014) has been written. Both are ready for review and pull requests have been sent.

Review complete. There are again comments on both the python-cpp-challenge PR and the tech note commits in GitHub. No major concerns here.

I know I said I'd try to reserve judgement on these tools until we started a broader conversation, but...wow. pybind11 looks fantastic, especially in comparison to Cython (which I might grade as worse than Swig overall, or at least no better).

Two issues (#176 and #179) are still open upstream with pybind11 and I will make some minor modifications to the implementation if these are fixed. But everything else is now working and ready for merge.

Note that this excludes the constness (as discussed before) and the extensions module which was out of scope. We may want to investigate the latter if a switch to pybind11 is considered.
My overall opinion is that pybind11 is by far the best tool I have seen for creating Python bindings for C++.

Got OK from Jim. Merged ticket branch of technote into master. Pull request for code will be merged by Jim into his private repository, possibly to be migrated to LSST repo later.

Upstream issues fixed. Apparently a nice interface for conversions did exist and I simply overlooked it in the docs. It was minor, so I modified it in code and technote. Now pybind11 looks even nicer! Wrapping up here.