Ent, thanks for all your work on this, and thanks to Demian and Martin for their reviews. In the meantime, the Python 3.5 release cycle has reached feature code cutoff so, at this point, generally only bug fixes are being accepted into 3.5, which means this refactoring will likely land in 3.6.  The next step is for a core developer to do a final review and then decide whether to commit it.  Berker has been commenting on this issue so perhaps he will have time to handle it as 3.6 gets underway.  (http is one of the modules/packages in the standard library where currently no core developer has volunteered to be its "expert" so we all share responsibility for it as time and interest permits.

https://docs.python.org/devguide/experts.html)