Here are my thoughts on Darkrai without the sleep clause. I shall refer to the
Ubers Viability Ranking Thread's tiers for the sake of some of my arguments, so look there if you are unsure of what ranking I am referring to.
Without sleep clause, Darkrai immediately becomes an S-rank threat, comparable to Extremekiller Arceus, Arceus-Ghost and Kyogre. It can sweep a significant proportion of the metagame with little support and is just as much of an overcentralizing force as those three, as every team needs to have a way to deal with it to succeed competitively. As others ahave pointed out, overcentralization is not really a concern for Ubers (especially as there are diverse ways to deal with Darkrai; it would not force everyone to run a certain Pokemon or anything of the sort). Thus, what I have mentioned so far does not inherently justify retaining Sleep Clause.
However, let us compare ways of dealing with Darkrai and ways of dealing with the other three major threats mentioned above. In order to find a truly solid counter to Kyogre, Extremekiller or Arceus-Ghost, one needn't look beyond B-rank Pokemon. Latias can deal with most, if not all Kyogre single-handedly, Skarmory and Terrakion handle Extremekiller, and Darkrai and Chansey take care of Arceus-Ghost. More importantly, however, all of these Pokemon can effectively perform roles beyond countering these Pokemon and are generally viable in Ubers.
Conversely, Darkrai simply has no solid counter among the Pokemon that are generally capable in the tier. That is not to say that it cannot be dealt with, but there is simply no Pokemon that single-handedly, reliably counters Darkrai, without resorting to Pokemon with Insomnia/Vital Spirit/etc, which are terrible Pokemon for anything else (and essentially put your team at a disadvantage against any team without sleep-inducers). Hydration requires rain, and none of its users are great, not even Manaphy. Prankster Pokemon, Scarfs, Mewtwo, etc don't want to switch into Darkrai. Sleep Talk is not reliable, and not too many Pokemon are happy to run the move. Lum Berry is only useful once (and only if no other Pokemon tries to status it) and is usually not the preferred item for a Pokemon. For all the reasons stated above, I argue that there is no good counter for Darkrai, the way that such exist for the other top threats.
In addition, as many others have mentioned, there are a lot of luck factors involved in the metagame as it is right now. There is a general consensus, with which I agree, that some degree of luck is necessary for an enjoyable metagame. However, sleep directly and indirectly brings in many luck factors that in no way make the metagame more enjoyable. Since no one really wants to run the terrible Pokemon that reliably counter Darkrai, and Hydration Pokemon aren't great either, most players are forced to run Sleep Talk to have Pokemon to switch into Darkrai (unless they want to run lots of random lum berries, which only work once and generally make that Pokemon less effective at its actual role). Sleep Talk is a gamble in itself. Moreover, as the number of turns a Pokemon sleeps is variable, untimely awakenings introduce another luck factor. Dark Void's 80% accuracy always existed, but is now much more significant, and can also work against the receiving player, as a choice Pokemon locking itself into sleep talk is set up fodder if Dark Void misses. The nature of sleep talk and sleep also introduces a lot of 50:50 predictions, which aren't great.
So, although I have done my best to be open-minded to the idea of a metagame without Sleep Clause and have considered the arguments of the few that have defended it, I will vote against it, for the above reasons. I feel that there is currently no good, reliable way to deal with Darkrai without weakening your team, and the best options (as in the ones that are least detrimental to your team when it is facing something other than Darkrai or another sleep inducer) to handle it involve too much luck for an enjoyable metagame.