Thanks for the responses, which are all helpful.
Oktas8 correctly summarises one of the points, which I made rather poorly in the original post. There were really two:
1) The timescale over which the decision to shutdown an engine is made should not be as compressed as it is in the EFATO drill.
2) Forcing the identification before application of power is permitted is unrealistic, and further compresses the timescale for decision to be made.
Blueline makes the good point that drills always concentrate on the worst case. I think this is an extreme though. If I have a partial engine failure in an SEP, I would adjust speed for minimum drag and assess the situation. I wouldn't immediately click into "land the aeroplane without power in a field" mode.
But the MEP equivalent appears to require a decision to shut down the engine very rapidly (5 seconds), or the test is "failed". While the application of rudder and the control of pitch attitude must, without doubt, be immediate, the rest can follow after a careful assessment based on all the available information.
FWIW, I have spent three years flying MEPs before doing this upgrade, and it's perfectly reasonable to suggest that partial failures should be covered in basic MEP training, leaving only the worst-case drill to cover in the IR. But more than any other drill I've come across in aviation, I get the impression that the speed required in this one teaches bad habits.
This is certainly not the fault of the FTO -- five hours to train for the upgrade is a very short timescale, and presumably this is the sames timescale as used in the full-course IR for EFATO and SE work. And I'm not blaming the examination system as such. I'm just making a case that a re-evaluation of the nature of the failure simulated in the test.
An anecdote: Some weeks ago I was practising ILSs in twin. A few seconds into the go around, after cleaning up. I experienced a sudden and unexpected roll and yaw to the left. This is the situation in which, within 5 seconds, I should be taking the decision to shut down the left engine. I'm glad I didn't -- the
problem was that the right flap had stuck down. Returning the flap to its extended position fixed the problem. Shutting down the left engine would almost certainly have killed me and my safety pilot.
Finally, I'm indebted to Stan Evil for setting me straight on a technical point that I should have known. I do, however, think that the drag offered by a stationary propeller even in a partially feathered state is rather less than the windmilling fully-fine prop that we're used to thinking about.