I think it's wrong to kill sleeping humans both because I'm often a sleeping human that doesn't want to be killed, and because I would see it as killing a (somewhat distant) part of myself. It's half "I won't kill you if you won't kill me" and half valuing the human gene code and the arrangements of thoughts that make up a human mind. I want humanity in the abstract to thrive, regardless of how I might feel about any individual part of it.

I think I agree with the bulk of the article you linked, but don't think I agree that it resolves my crux. To quote its quote-

This intuition is epitomised in Jan Narveson’s (1973) statement, “We are in favor of making people happy, but neutral about making happy people.”

I do not believe we are obliged to create entities (be they humans, cows, insects, or any other category) but we are not obliged not to do so. I think we are obliged to avoid creating entities that would prefer not to have been created. That still leaves us the option of creating entities that would prefer to have been created if we want to- for example, nobody is obliged to have children, but if they want to have children they can as long as they reasonably suspect those children would want to have existed. If I want cows to exist, I can morally cause that to happen as long as I take reasonable precautions to make sure they prefer existing. As you say, they might well not want to exist if that existence is being locked in a box and hurt. As I say, they probably do want to exist if that existence is wandering around a green pasture with the herd.

I'd like to grab an example from the article you linked, the one about the Buddhist monks in the collapsing temple. As it says

Imagine a large temple filled with 1,000 Buddhist monks who are all absorbed in meditation; their minds are at rest in flawless contentment. Unfortunately, the whole temple will collapse in ten minutes and all the monks will be killed. You cannot do anything to prevent the temple from collapsing, but you have the option to press a button that will release a gaseous designer drug into the temple. The drug will reliably produce extreme heights of pleasure and euphoria with no side effects. Would you press the button? [footnote: Alternatively, to avoid potentially distorting ideas about, for example, violating their autonomy: Would the accidental release (by environmental forces) of such a gaseous compound make the world better?]

...

It is tempting to feel roughly indifferent here: Pressing the button seems nice, and assuming it produces no harm or panic in the temple, it may be hard to imagine how it would be something bad. At the same time, it does not seem particularly important or morally pressing to push the button.

This is what I was trying to get at with the usage of "suboptimal" above. If I'm going to encourage the creation of cows for me to eat, I'm obliged to make their existence generally positive, but I'm not obliged to make that existence euphoric. Positive sum, but not optimal.

While a world where one's life and death is one's own choice is a good world in my view, I can't find myself getting axiomatically worked up over others acting on my behalf. I'm willing to make acausial deals- for example, if a man found me collapsed on the side of the road and only took me to the hospital because he figured I'd be willing to pay him for doing so when I woke up, I'd pay him. I prefer a world with a strong authority that promises to find and kill anyone besides the authority that kills a human to one without any strong authority, and even though I would try and escape justice in the event that I committed murder I wouldn't argue that the authority was doing anything immoral. (Both examples have details that are worth adjusting, such as how much I'd be willing to pay the rescuer or mitigating factors such as manslaughter vs first degree murder, but my reactions to which seem to point at different values than you.)