If porting Sigil's QtWebkit dependencies to QtWebEngine were trivial (or if trading QtWebKit's bugs/deficiencies for QtWebEngine's bugs/deficiencies resulted in an overall positive gain) we would have already done it.
That Cmake can successfully generate Sigil makefiles (or VS projects) with Qt5.7 is no guarantee that smooth sailing lies ahead. I know from personal experience that it doesn't!
It makes far, far more sense to custom compile Qt5.6/7.x with QtWebKit added back in (which is fairly trivial--even on Windows) and use the time gained to determine if a port to QtWebEngine is feasible, or even wise in the long run. No modification to Sigil's code is necessary for that route.