The ballot isn't about "realistic smash picks". Like, ever. It was about what people were interested in.
More ideally, yeah, people would just vote for what they want and not care about likelihood. But in practice, perceived likelihood generally affects performance in polls and we saw it with the ballot, where for example AT'd characters like Waluigi or Ashley got less traction than they otherwise would have while non-AT'd Isaac had an easier time coming to the forefront, and generally Microsoft/Sony characters or others who weren't on Ninty consoles at the time didn't do as well as they would, say, these days. It's easy to forget it now after he's gotten in and been championed as a fan pick, but Ridley absolutely got overshadowed by some of the other big fan picks like K. Rool, Isaac, Banjo, etc. in the aftermath of getting boss'd in 4. He still did well on the ballot (because as you say some fans will vote regardless) but it probably wasn't commensurate with his actual popularity standing. Basically all of the data and anything you can glean from what people were putting out on the net regarding the ballot, even in more "mainstream" media outlets, I think is consistent with this.
orget the Smash bubble. Tons of iconic characters and franchises get votes simply because the franchise exists. Remember, we're the minority, the hard core fans. The casuals were the biggest turnout. That's the reason why many votes for ones like Goku etc. because they don't care about any kind of random "appearance on a console" or whatnot. These aren't important factors to casual players. Do you honestly think they care if it's plausible? No. Because it was made clear Sakurai had to actually ignore votes because he knew they would happen. Plausibility isn't as important as you think it is. Especially when people weren't realistically expecting all characters back, and Pichu wasn't exactly expected as having some yearning for its return due to being fun to play as. Young Link? Definitely not a popular pick for returning among the hardcores. Toon Link hit that niche(though not the exact moveset style) as is, so at most said fanbase was good with switching them back and forth so the sole "child Link" niche was kept, with other hardcore fans wanting to keep it to one Link, regardless.
What you're saying about the Melee vets and the hardcore minority vs. casual silent majority is actually contradictory. Votes for the Melee vets showed up in the exit polls significantly; a lot of hardcore fans also subscribe to vet pandering and it showed.
Perceived likelihood is far from the only factor, and in many cases I don't really like it being a factor, but as I said basically any tangible evidence you can find points to it playing a significant role in this case. There were some exceptions, like Banjo or veterans who certain circles badly wanted back, for example, which had
specific reasons for getting votes despite seeming nigh impossible. But broadly the perceived likelihood factor had its greatest effect I think when a character was viewed not simply as unlikely, but rendered logistically or systematically impossible to some (i.e. people not believing in AT/boss promotions, people believing in the Nintendo platform appearance rule, etc.).
I acknowledged the possibility of a large, silent casual voting audience going overwhelmingly towards certain picks, particularly various non-Ninty picks that were viewed as impossible by much of the hardcore fanbase, but there's...basically no evidence to suggest that played a significant role. It seems equally likely without further evidence either way that the body of casual votes was too split between random characters to push any particular pick like a Terry or a Hero towards the top. Sure, those characters made it into Ult, but as has been acknowledged, those characters simply had other reasons besides ballot votes that made them fit the bill for a DLC slot.
Of course plenty of franchises got votes simply for being major players in general gaming, but I don't think it was nearly as many as you purport, since the ballot was advertised and circulated primarily among people already playing Smash, already owning and being invested in Ninty platforms/games. At that time, picks like Chief or Terry had little to no actual traction in any visible circles, and it's hard to imagine a cluster of casual votes that were focused/united enough to push any one of them to the top. Who knows, maybe the casual audience went overwhelmingly for one of the picks that intersected with the hardcore votes, like a Bomberman for example. Seems just about as plausible as them going for someone like Terry (sorry to pick on him like multiple times).
The idea it did poorly has no legitimate founding
What is the legitimate founding for it doing well, though, besides just "it's a mega-popular franchise"? Part of the point is that plenty of mega-popular franchises simply didn't do well because not all fans of a given franchise also want that franchise in Smash (or care about Smash at all), or view it as a particularly reasonable newcomer request. Some will ignore the "reasonableness" of the request, but some will not. The only evidence of MC doing well is simple guesswork that a popular franchise probably got some votes. To the contrary, the polling data and general activity level surrounding the possibility of MC in Smash at the time suggest that it didn't do as well as its popularity among non-Smash fans might suggest.
Also, we know it was in talks around 2015, right after the ballot ended.
It is not confirmed that it was after the ballot ended. In Source Gaming's "Smash Ultimate Development Timeline," one of the linked sources suggests that negotiations for Steve could have started as far back as 2014, before the ballot. In which case the ballot would have had nothing to do with Steve's inclusion.