Now please don't get me wrong and I'm not a saying it's a brilliant idea or that it could be ever economically justifiable. But given the right mix and using proper techniques it does work very well.
In the early 90's we did in fact gain FAA approval for single engine operations on the Tridair's Gemini 206L-1. 9 (Covered in supplement 6 of the Gemini flight manual dated 28 July 1994).
It was the right mix as we were trying to produce the ultimate Cat A helicopter, in that an engine failure on take off had very little effect on the twin engine performance. We had two 450 shp engines and the transmission was rated at 456shp. The systems were developed so that the loss of the engine after decision point,
full power was immediately available with out exceeding any engine limit.
Returning from Heli Expo in Miami 1992 we always took off on two and shut down one engine when in cruise. The big impact here was that the C20R at MPC cruise used quite a bit less fuel that the original C28 or C30 at the same power setting.
By alternating the engine shut down on each leg each engine had less time on them than the airframe. Miami to Seattle via LA it made quite a difference.
We always started the "spare" prior to landing and it really was a easily managed process. The technique used to prevent clutch grab at the restart was to beep the engine down prior to shut down so the it was completely off line. We use triple needles on the torque indicator one for each engine and one for what was going into the transmission. So in cruise it would display 85% at the transmission and 42.5% for Eng 1 and Eng 2. As the engine was beeped off line the torque on that engine was drop to 0 while the running engine would rise to 85%. When restarted we would get it running to flight idle delivery zero torque then slowly beep it up to 42.5% while doing this the running engine would drop also to 42.5%.
Bell installed our kits in the Twin Ranger version of the L4, while being an excellent Cat A performer they did not want the single twin option.
It's kind of interesting for us to see the recent announcement mentioned above, more than 20 years after our experiments. Could have saved them some money perhaps.
