At this point, I have to say that I'm reasonably convinced the leak is true just by there not having been any damning evidence against it after this long (against the CSS/SSS, at any rate; how the rest of the stuff relates is a mess that I've long since stopped trying to sort through and I'm pretty sure is severable from the meat of the leak anyway).
It's funny, because the one point that most seem to claim most vehemently being for it, that being the Tomodachi Life stage icon, doesn't really carry any weight in my opinion. I'm more impressed by how believably the content we knew about already is being represented on the CSS. The fact that the CSS renders for Peach, Rosalina, and a number of others were present for a split-second in that 3DS modes trailer would have been super-easy to overlook (indeed, I'd plumb forgotten it and Rosalina's render seemed strange because of it, but matches up perfectly in hindsight); Fox's slot color changing being corroborated elsewhere (although oddly enough that's not depicted accurately in the second post), etc. I don't know that most have an issue with this specifically at this point, but of course the names would've likely been removed if the full 49+ characters were going to show up on one screen; at a resolution of 400 pixels, that leaves a mere 45 pixels per name (that's 4 pixels per character in "Meta Knight"), which would take up valuable screen space whilst being completely unreadable. The interspersing of old and new stages does seem odd, but the selection itself seems pretty sensible (though I'll admit that might be partially personal bias speaking, as WarioWare and PictoChat were my favorite non-neutral Brawl stages). And I know this concerns the Ninka-Vaanrose leak more so than this (they seem pretty intertwined anyway), but I can't come up with any way anyone should have known about the new alternate colors for Mario and Bowser prior to E3, so I'm convinced there are legitimate sources related to this somehow.
Really, with the amount of time this leak has stood at least viable with pictorial evidence (which is inherently very disprovable), it sickens me to hear this repeatedly compared to the Gematsu leak. Not just because I never bought it 100% (admittedly, with Shulk's apparent inclusion, I'm more willing now to concede that there might've been some legitimate information from early in development behind it), but because that list of characters, no matter how many were correct, could not have possibly been disproved until launch without explicitly deconfirming one of them. As Das Koopa said a bit ago, it operated on a different timescale entirely, namely the speed at which characters were being confirmed. I'll extend that and say that text leaks are ALWAYS going to operate at a ton slower of a timescale, as there's only as much detail in a text leak as is explicit in the text, whereas a pictorial leak of anything mildly interesting has a number of things that could reasonably disconfirm it even if they weren't the main focus of the picture.
But that's all just my two cents; I'm still not remotely close to 100% sure about the leak, but I'm eager to see how its likelihood'll change based on what we see from future P's-otD and so forth.