MC-240030 was fixed by making it so that any chunk containing at least one bedrock block at Y=0 will have new generation placed under any column where Y=0 is a non-air block. This fix, however, has a minor flaw, as noted in the comments of MC-240030: it is possible for gravel to generate at y=0 and then spontaneously fall, resulting in air existing at y=0 without player interaction. When the world is upgraded, large columns of air result which stretch down to the void layer.

I'm not too sure as to how this issue could be fixed - perhaps instead of checking only Y=0, it'd check if all of the blocks from Y=0 to Y=5 are air instead, and only then would it not generate new terrain under the world? This would fix the issue for 99.99% of cases where this falling gravel edge case would arise.