This is now much easier than before!
Some may remember my old thread where I and a few others stumbled through this process. One of the most vexing aspects of that whole thing - the one which almost made me give up before I'd really even begun (ffmpeg) - is now completely unnecessary.
Here are the current, tested steps:
1) Go into SE settings > Display and change the view from "perspective" to "cylindrical".
2) Set the windowed screen resolution in SE to 3216 x 1608 for a resulting 4k 360 video.
3) Export MP4 video.
4) Inject the 360 data into the resulting MP4 with the Spatial Media Metadata injector tool.
5) Upload the tagged 360 MP4 to YouTube
6) Download the full-resolution 360 encoded VR video when it's done processing.
Unfortunately even though there are plugins to load and edit all sorts of video files, VirtualDub can only append direct streamed AVI, so if you're wanting to attach several video segments together you'll first have to turn those back into AVI format, append them, and then use ffmpeg to turn that entire file back into an MP4 once again. This is kind of a pain, but the advantage is that you can much more easily process a single file now.
Here is an example of a full-res VR video made using this method, and SE v0.991 beta: