These radar altimeters have very wide front-ends, as as a result are highly susceptible to out-of-band emissions. The radars are allocated from 4.2 to 4.4 GHz, so the bottom edge of the radar altimeter band is 200 MHz above the top end of the 5G band in question (3.98 GHz).
The fact that these radar altimeters are susceptible to out-of-band interference that is literally further away in the spectrum than the width of their entire allocation is IMO an engineering failure. Yes, spectrum management (rightfully) tends to err on the side of not forcing the incumbents to make modifications to existing systems. However, better filtering would resolve this issue and it is shocking that the engineers who designed them didn't foresee that other services would eventually be allocated in adjacent bands. Avionics is often NRE dominated, they couldn't have sprung for a cavity filter? Or if that was too expensive, designed more headroom in the LNA before saturation and filtered the out of band signal with a crystal or SAW filter at IF? Rejecting out of band that far away is not rocket science.
I don't think it's unreasonable in this case to temporarily suspend the rollout and give the aviation industry a deadline to fix its radars, but it should only be temporary. Any other radio service would be told to take a hike if it keeled over due to out of band interference that far away.