I've been chatting frequently with the Ansible team, and an initial plan exists to reshape the work in preparation for it to reach mainline, but doing this in a long-term maintainable manner will require a little design and refactoring of some parts of the Ansible code, as well as redesign of the Mitogen integration.
Already as a result of these chats, the Ansible team have supplied a proven solution to one of the most difficult aspects for completing Mitogen – Python 3 support. I hope this small example can become a model for future cooperation.
My current plan is to complete the extension using the current design, so
that existing deployments can be supported without upgrading Ansible, while beginning to catalogue all the changes necessary for an eventual mainline integration - https://github.com/dw/mitogen/issues/135
In the meantime, there is significant benefit to developing the extension out of tree, not least the ability to target existing deployed Ansible versions. Much of the work done on the extension is experimental, and there isn't much existing design precedent to borrow ideas from. Chances are high that a few more fundamental changes will be required in the extension before its design fully settles – that kind of work is much easier to iterate when it is not directly integrated into an existing stable product.