Unless Mono can get the tech spec out of the workshop, or someone else has the equivalent I suspect you won't get a definitive answer. But as this appears to be a crew room discussion, rather than an operational question, I'll throw my general opinion in. Hopefully it will help you construct an informed argument for your friends.
A radalt is a radar that tracks the ground, the altitude is an output from the range gate. If the radalt loses lock, for whatever reason, then it will declare unlocked and attempt to re-acquire, this could cause the output to become NCD. Similarly some radalts calculate a track quality, and decalre the data as NCD when the track quality drops below a certain threshold. Most radalts have a memory flag to avoid going NCD whilst attempting to re-acquire.
In level flight, over perfectly flat reflective ground, then the ground speed would have little effect. But as this is not the case then we have to consider it. The radalt will have a large footprint in an attempt to even out scatter effects, and more importantly detect a leading edge from the closest point of the ground under the aircraft. It will also, probably, have a means of eliminating doppler effects from the calculation. Amongst the possible solutions is using a saw tooth waveform and calculating on both the up and down slope. As such there are limitations to the processing. Ultimately the dominant factor will be the velocity along the radalt's range track direction. As the aircarft pitches up the radalt will begin to see a significant component of the groundspeed along this axis. Although the actual range track will stay fairly constant as the distant to the ground's closest point remains steady. Without knowing the details of the tracking algorithm it is impossible to say when the radalt will lose track. Similarly without knowing how track quality is used, it is impossible to say at what point that would cause a loss of data output. The TSO points to a set of max orthogonal velocities and a max attitude of the aircraft for which the radalt has to keep lock, and retain sufficient quality of track to meet the accuracy requirements. Outside of that I'm afraid I can't tell you what would happen.
In summary it is not so easy as saying at groundspeed x or vertical speed y the output will become invalid, there is a lot more to it than that, including aircraft manouevre and nature of the terrain. As I have said to my colleagues recently, it is all about when the radalt loses track, and it is wrong to assume that at a certain speed radalt data will automatically show as NCD on an FDR.