Of course, its only a private build and I would never expect this to get accepted into current Stretch repo. The idea is to move the build from 4.8 to 4.9, with a few improvements, but keep the package partitions, time stamp, and build trees the same.
I've had good results with my NMU build and I use it daily. When I see the effort the Debian qemu team has made to keep qemu-2.8 current with Security fixes my confidence is raised too. I put a hold on the associated qemu-2.8 build (using libxen-4.9 headers) with `apt-mark hold xxyy`.
However, looking at how Buster packaged Xen with qemu makes me think that now and the next 6 months is a chance to change the direction for the better, moving Xen from 4.8.1 to 4.10.3 (eventually in June 2019) and build qemu with libxen-4.10 headers and libs.
Unfortunately, Buster has chosen xen-4.8.1 with qemu-2.10 for now, although I see no reason this choice can't change.
When xen-4.10 is released is when I will start with this project, unless Debian developers start first. There is no reason for me to build or test 4.10 in Buster now. I'm doing this in Ubuntu Artful.
If anyone is interested in any part of this, support or criticism, join us in