Hopefully subject is as clear as possible :-)
The benefits of this include
* more stable browser tests that the VisualEditor team will be able to keep an eye on that don't hide false positives in our own product.
* smaller extension for third parties who don't use VE
* less bugs in VisualEditor tablet software due to our code clashing with VE code.
* no tests filed against the mobile web backlog which would lead to a cleaner Phabricator board
* research through doing about how mobile can use oojs ui and identify the problems we need to solve to do that.
To do this I would suggest the following steps (which may need fleshing out and may uncover additional work/spikes)
* rewrite the header of VisualEditorOverlay as oojs ui components
* rewrite the edit summary pane as a Oojs ui component
* rewrite VisualEditorOverlay as a oo.ui.ProcessDialog
* Move all VisualEditor code to VisualEditor extension - browser tests, js, css, RL modules