I don't think he's suggesting that.
Consider he gets a very stable approach - either multi engine, single engine or asymmetric - matters not. Stable approach.
Gets to MDA but is still a way from the threshold and slant vis does not allow sight of approach lights, threshold, whatever.
One of his instructors says immediate go around, the other instructor says fly level at, or up to 50 feet above, MDA until the MAP. IF DURING THIS LEVEL SEGMENT he gets the required visual reference(s) THEN he makes the decision to continue THAT approach in a stable manner OR carry out a visual circle to land (still stable).
Nobody has suggested diving to the surface in either of his questions.
I think the second question is more along the lines of - theres a required "altitude at distance" point of, in his example, 1200 feet agl at 3 dme. The approach plate shows the FAF as 8dme - but a 500 fpm constant descent will not get him to 1200 at the same time as getting to 3dme.
So, does he (a) delay TOD until a 500fpm WOULD get him to 1200@3dme, (b) decsend at a lower fpm to choreograph arriving at 1200'/3dme, or (c) start the 500fpm at FAF, then level off at 1200 and fly level to 3dme before starting a second, stable, 500fpm, descent (presumably 3° slope on the IR flight test as suggested).
No diving - we all know that's a bad thing.
Delay TOD, descend gently, or level off before 3dme and descend again?