There's no reason whatsoever to fly the VD manoeuvrem ever. It's a theoretical design case and can (and should) be done by simulation/analysis/prediction. (With the data used then validated by various other means)
For the 253 manoeuvres (upsets) you obviously have to fly the upsets as specified. (Though if you know which one or ones are critical, just do those of course)
For the purpose of expanding the flutter envelope you can and should fly the manouvre in a suitably cautious manner. I would say you fly it definitely no more aggresively than the 335 case, because that's the one you designed to. But since all that really matters is achieving the speed, a more shallow dive to the same speed achieves that goal and is safer with regard to pulling out of the dive if a problem is seen.
If you already have the aircraft certified and you're doing a delta, you have two options - if there's no threat to the upset cert, then you can do the "flutter-type" dive to a target speed to re-establish a test point at Vdf/Mdf with your new config. If the upsets need to be re-examined, what has been found acceptable is to repeat the upsets with the new config and take credit for the speed achieved in terms of showing freedom from excessive vibration (251). While you won't hit the exact same speed, you'll be close enough to demonstrate the requirement effectively. (If the original type design max upset speed was 0.88, and you get 0.875 in a "repeat", that's the same speed for all practical purposs, given variability in the test)
These are not low risk manoeuvres, it would be (IMHO) irresponsible for the applicant to conduct these repeatedly in an attempt to perfectly reproduce a previous result, and equally irresponsible for the authorities to insist on the applicant doing so.