All this talk about whether LNAV is engaged, is all a bit irrelevant if as I suspect the autopilot has failed, due to the flight controls degrading to secondary.
As GBO correctly concludes, the analysis of the primary returns coming back to Penang show no straight line portions… after the turn at Penang and after 18:01z we have no radar returns to work from. Therefore, we don’t know if LNAV was engaged or not.
I see no reason to make the claim the autopilot was ever engaged again and connected to an active route….
GBO is correct in his initial failure scenario, it’s just the aftermath of that failure that need to be ironed out.