In hindsight, I think my original post was a bit hasty/not nearly thorough enough. Probably because most people I've talked to about this already were also against Sleep, and I wrongly assumed that the vast majority of people felt this way. Let me rectify this by building a more exhaustive case.
1. Sleep Is Uncompetitive.
This is a fact that the community has already widely agreed upon. We know this because of the existence of Sleep Clause in the first place. We play with a modified version of the game where Sleep is severely nerfed compared to its potential. Obviously, this isn't a very useful point on its own since we do play with Sleep Clause. I brought it up because it's relevant to my guiding philosophy: if issues arise with any Pokemon due to their ability to take advantage of a heavily restricted Sleep, we should look at the mechanic itself rather than the Pokemon. I've seen a couple of people say to just ban Darkrai or Iron Valiant, but I would ask: why preserve a mechanic we already bend over backwards to allow in a competitive setting instead of just getting rid of it?
With all that said, the real question to answer is: "Is Sleep with Sleep Clause active uncompetitive?" To this, I would argue a resounding yes. The amount of RNG introduced by Sleep is a detriment to the meta, both in terms of Sleep moves themselves and the accuracy of the moves. I think the best way I can explain this is by replying to this post:
I don't mean to single idkup out in particular, but I've seen this thought process a few times and I believe it's missing the forest for the trees. It's precisely
because things like resist berries, Custap Berry, Rowap Berry, and Red Card are reasonable/expected as counterplay where Sleep is too much. Let me illustrate this with an example that's very common: Iron Valiant vs Galarian Slowking. The most consistent counterplay to Valiant this generation has been a Poison-type with a resist berry, and Glowking will most commonly run Colbur vs Valiant to guard against Knock Off. To get around this, Valiants can run Ghost-moves instead. But then Glowking can also be Kasib if it wants to. Why not just dodge the mindgames entirely and put Glowking to sleep on the switch with Hypnosis, though? If you go Lum, now you're vulnerable to standard coverage. If you go Sleep Talk, that's a massive commitment for something that might not get any value at all, EVEN IF the opponent brings Hypnosis.
I don't think Sleep is healthy precisely because of this ability to bypass common draft tech through random chance. One turn of doing nothing can certainly lead to instant loss in SV. However, this is always in the Sleep move user's favor to enable. Sleep gives a guarantee of one completely free turn to do whatever you want, provided you hit. If you miss, the power swings in the opponent's direction, in theory. This swinginess determined by a Sleep move hitting or not is degenerate for a competitive meta. Also, the two most notable Hypnosis users, Darkrai and Iron Valiant, are Pokemon that force many switches, and using Hypnosis on the switch is hardly a risk at all.
2. Sleep Clause Is Not Cartridge Accurate
This part of the argument is more philosophical, but it was very relevant when OU went through this same debate a while ago. As currently implemented, Sleep Clause is unable to be replicated on cartridge. This is concerning, since the game we play is on a simulator that's meant to be accurate to cartridge gameplay. While it does originate from Pokemon Stadium, it has never been a part of the main series games. The closest you can get to Sleep Clause on cartridge is agreeing to not use a Sleep move again after putting a foe to sleep until they wake up or faint. There are a few situations where the optimal play is different depending on whether Showdown's Sleep Clause mod is enabled, or you are playing with the gentleman's agreement version. A simple example is wanting to put a foe back to sleep on the same turn that they wake up, but if your opponent switches out, you would violate "Sleep Clause" on cartridge by making this play. On Showdown, you're free to spam all the sleep moves you want and are predicted by Sleep Clause in case of any wrongdoing.
How much this matters to you is dependent on your personal philosophy of the game we play. I personally believe we should strive to be as cartridge accurate as possible, so I don't like Sleep Clause on principle. The mechanic is clearly not balanced enough to allow with no restriction, so at that point, I would rather get rid of it than play with a mod of the game.
3. Sleep Is on the Rise.
While I don't have any tangible evidence to base this on other than looking at the raw replay data of DCL and anecdotally seeing Hypnosis way more frequently than years prior, I believe the usage of Sleep moves is just going to get higher as time passes. The main idea is that you can use Sleep to try to overcome a losing matchup through RNG, or you can even bring it in a good matchup where the risk for missing is relatively low, but the reward could win you the game on the spot. When this is more understood, Sleep usage will increase, and we are already seeing it on more than just the best users Darkrai/Iron Valiant.
4. Why Bother? Is It That Big of a Deal?
Admittedly, no, Sleep is not a "major" issue concerning SV draft...right now. I believe we have the power to nip it in the bud early, though. The future doesn't even matter, however, since usage isn't inherently linked to whether something is healthy in the meta or not. On a fundamental level, Sleep is uncompetitive and requires a game mod to be usable. We have fully banned things for lesser offenses (King's Rock, Quick Claw) than Sleep. Just because something isn't overtaking the meta doesn't mean we can't remove it if it's still a net benefit. A slight improvement is still an improvement, after all.
5. But What About Yawn?
As already pointed out by
scionicle , Hypnosis is significantly more common than Yawn in DCL at this point. I can contribute to this stat too since I also brought Hypnosis on Darkrai one week where I did not click it in battle. In an ideal world I would also like to preserve Yawn in some way, but I am more than willing to lose it as collateral since it does not have a very relevant impact on the meta. Almost all of the relevant Yawn users have some other form of status, disruption, or slow momentum to use instead.
-Galarian Slowking (Has Thunder Wave/Toxic/Chilly Reception)
-Slowking (has Thunder Wave/Chilly Reception)
-Uxie (Has Thunder Wave/Encore/U-turn)
-Blastoise (has Flip Turn/Roar
-Swampert (has Flip Turn/Roar)
-Empoleon (has Flip Turn/Roar
-Torkoal (would be hurt the most by it, but still has the niche of Sun and Ninetales is running Hypnosis anyway so...)
Some of the mons on this list have dubious draft viability already, so any mons not listed here are even worse. Losing Yawn hurts them, obviously, but the beautiful thing about draft is we can decrease their prices to reflect this. Since Yawn clearly isn't being used much already, this could even be a positive for some of them!
6. Sometimes the Status Quo Is Wrong
If Sleep were introduced in Gen 9, or even as far back as Gen 6, there's not a doubt in my mind that it would be banned outright. Sleep Clause is archaic and would never be created today with modern tiering philosophy. This is taken straight from the OU debate, and that's on purpose because I think we are following a very similar trajectory that OU did. Sleep was never an issue until Darkrai/Valiant came around, nothing happens at first, but then it slowly starts to become more popular and resentment grows until it finally becomes too much of an issue to ignore. I believe we are currently at that second to last stage where resentment is growing. I can offer no proof other than I feel I'm getting deja vu from the OU metagame discussion thread about Sleep.
7. "Just Ban Paralysis or Ice Moves that Freeze Too Lol"
Obviously, Pokemon is full of RNG and variance and this is unavoidable. The difference between Sleep and the other statuses is it's actually within our power to do something about it, AND WE ALREADY HAVE! By playing with Sleep Clause, we have already collectively agreed that Sleep is more uncompetitive than Paralysis, at the very least. I'm sure we'd all love to get rid of Freeze as a mechanic, but it's just not possible. We draw the line somewhere with RNG (OHKO moves, evasion, King's Rock), and I argue that Sleep, even with Sleep Clause, is past this line too.
8. Conclusion
Sleep is inherently uncompetitive and introduces unhealthy variance to a meta that is already volatile. The cherry on top is Sleep Clause, as currently implemented, isn't even cartridge accurate. I see no real merit in keeping Sleep around, especially since Yawn is basically irrelevant. What separates Sleep from other sources of RNG is our very realistic ability to take action on it.