Technically, both are allowed. But that's not the real question here. The question is which scenario would be chosen in a Rawlsian Original Position.
To decide that, I would argue, there isn't enough information. That is, Nozick's claim is unjustified and probably mere polemics.
1. Distribution vs. political system
What we are given are allocations (allocation is the distribution of economic goods) of a single good, of monetary wealth. But that is not what Rawls' Theory of Justice is about. It is about a fair political system, including socio-economic rules. This includes themes like inheritance, education, and access to health care and economical, social, and political positions. Thus, taking wealth as singular measure is misleading:
It presupposes the premise of economic utilitarianism: wealth is equal to happiness/goodness. That is an undue restriction of the concept of good at hand and also ridiculous considering that Rawls explicitly tries to argue against utilitarianism as a valid measure of justice. And there is nothing in the difference principle that demands this reduction as well.
2. Stable system vs. one-off moment
In the Original Position, we have humans who do know how economic systems, motivational systems, and moral intuitions work like. They may be thought to be behind a thick veil of ignorance in the sense that they do not know where they, individually, will end up in terms of gender, abilities and socioeconomic status. But they are not dumb.
Thus, when they decide about how much inequality should be allowed (even if they only look at wealth for a moment), they do not look at where they may start. Mind, they don't even know in which "era" or "generation" they end up. They look at where they may get. That's why access to education and positions is that important in the theory.
It is also the reason why a more unequal but equally stable economic system which allows for a rapid positive development of the minimum will IMHO quite obviously be preferred over a stable, but stagnant system (ceteris paribus, ie. all other things equal):
Assume that scenario/system D allows for every group to be at 25% more every five years and in scenario/system E the LAG will be at 10% more while the others gained 50% in the same time frame. Now, guess which one is judged as better (ie. more fair) according to the difference principle. It will be D, even though the overall wealth will grow slower than in E, since the growing inequality is not justified by comparable growth for the LAG. That is indeed contrary to what economical utilitarianism would suggest, but it simply is more fair.
Summary
Although allocation scenarios like the one in your table are common to illustrate the difference principle and how it interrelates with the minimax principle, especially in economics, it is misleading: Systems are dynamic, they are no mere distributions. And a system is fair if it systematically produces fair outcomes over time. What Rawls has in mind is how to shape a system that involves all socioeconomic dimensions and can be considered as fair by impartial judges. Reducing that to one-off evaluations of wealth is quite misleading and ...well...unfair. It cannot possibly do...justice to Rawls' theory. Now I'm out of puns.
