It is reasonably common with Citrus for the scion to die but the rootstock to survive. This could be caused by a freeze but also other stress like drought or disease.
Sometimes the scion will die right back to the to the graft or bud and can recover if rootstock suckers are stopped from taking over.
Death beyond the graft is a loss of the scion variety. You could regraft the rootstock, but for rare scion varieties they may be hard to get again ?
It has been suggested that the graft union acts as a permanent cincture, and this is part of the reason grafted plants flower and fruit early.
( Apart from the maturity of the scion). As you say never 100% fully healed in that respect.
Plants also exhibit fungal dieback. A fungal infection causes progressive plant death down the stem to a lower point.
It is a strategy for the plant to escape the infection. Possibly a scion could be subject to this until the graft union is reached.
With Citrus I have often noticed that if suckers are allowed to grow well above the graft union, they can not only dominate due to vigour, but send the scion backwards and kill it eventually. ( Delayed rejection ? )
Overall I would say the closer related the rootstock and scion are related / the less incompatibility, then the more likely that both will behave similarly.
I certainly have some grafted Fingerlimes that have died back to below the rootstock and some seedling Fingerlimes that died back to the roots but resprouted, so I still have that variety.
Also I have quite a few interesting Citrus seedlings that I grew, where the original seedling was weak and now only survives as grafted plants that I propagated for insurance.
I think you would have to also consider the advantages and disadvantages of grafting to rootstock vs airlayers or even cuttings.
For rare plants you don't want to lose, or can't replace you may want a few versions as a back up plan.