This doesn't necessarily strengthen the argument. We don't know how much weaker the correlation was. The difference could be negligible, or it could have been caused by some other factor. For instance, if the studies including insomniacs were comparing them to people who for some reason were very resistant to melatonin, then the correlation between melatonin and inducing sleep might naturally be weaker, even if melatonin significantly helped the insomniacs.

Irrelevant. We don't need the sample to be representative of the entire human population, just of the subsection of that population who have insomnia.

This strengthens the author's conclusion. This tells us directly that when people with insomnia were studied, they derived no significant benefit from melatonin. This sugggests that melatonin isn’t very helpful for people with insomnia, which strengthens the author’s claim.

Irrelevant. Knowing about certain people in the control groups doesn't tell us anything about whether or not people with insomnia would benefit from melatonin.

We don’t know if anyone who has insomnia was actually given doses of melatonin in these studies. Thus, without further information, we can’t say whether this statement even applies to the argument.