If there are any mistakes in this reply, I defend my case that it's past midnight at the time I'm writing this and I'm somewhat not thinking properly but I still want to type this.
I am not in particular rejecting any of your ideas, but these have been discussed quite a few times through the community and the first one (about sound length) isn't gonna come any time soon. You can take this with a little grain of salt, if you want, I am merely commentating on this post with what I've seen through chat logs.
If you haven't noticed already, the original instruments and the 8-bit instruments are implemented differently. The original ones are just one shot audio files of the instruments and don't hold any particular length. Best reason to consider why Jacob did this is to keep the bandwidth of the site low, faster loading times, etc. Longer audio files that are uploaded on the site will obviously take more space. Better server to maintain the site also means higher monthly payments, I think Jacob mentioned this somewhere but I can't find the original message right now. I'm saying this by what I remember don't kill me if I'm wrong. EDIT:
https://onlinesequencer.net/logs?id=69352#m69352
The 8-bit instruments are created from web API resources, and the reason (?) why they have note length is because the soundfont is already existing on a different website, and OS utilizes the resource from there. I can assume that the resource isn't uploaded locally on OS, therefore not taking up any space. The only problem with the 8 bit instruments is that it can be super harsh for people with poor computers, or that it doesn't work on certain browsers.
In the end, trying to smoothen out the original instrument's "end of the note" after being held a certain length sounds hard to implement. The loop idea that you have might not sound bad, but this also means sacrificing some attack of the note (the bite that it gives when you press it down.)
As for the popularity suggestion, we actually do plan on getting it in the near future when Jacob isn't busy with work/too lazy to program the site. We've considered the like/dislike function as well as a five-star system, maybe trying to give commenting on a sequence a try. Things such as sequence rankings and composer points have been suggested as well, if that piques your interest, but I have no idea if those will come, ever. I have no idea what to say for the algorithm when the site stops deciding to consider something "popular," but I agree it should be implemented so that older sequences don't get an unfair advantage.
I apologize if I have gone off-course with your suggestion, but I hope this sheds some light on any possibilities bound to come. I guess I'll try to get Jacob to respond to this because he knows a hell ton more about his own site than I do. I just overlook the community almost every hour haha, so I see things like this pop often and I'll be happy to give out my thoughts on them.