At this point it is what it is. I can take away markdown support, or we can live with the minor edge cases. I honestly have no idea how many people are using markdown, I can't go by any data, all I can say is I feel like the downsides are all navigable. Nobody seems to be actually citing footnotes in their posts on a regular basis, and if they are we have lovely support for superscript tags and they work in the editor interface, so I don't think it's a big deal.
Honestly, in my opinion?
Given how any
:/ in e.g. typed https:// URL rendered as

(or whatever emoticon it was) previously (before you very quickly removed it, kudos!) I'd be all for taking down
all emoticon markdown myself... ;-) ;-/ ;-\ and similar work just fine, if not actually better.
Everybody understands ASCII emoticons anyway, they don't break line spacing (as the below
xxxD:
example, which renders at 1.5 lines in some browsers) and for the cases they aren't sufficient, Ars still supports Unicode emoticons (or rather translates them to inline pixel image ones, I believe).
Otherwise you might run into more and more edge cases that break things, IMHO...
Case in point:
PMID: identifier in scientific citations renders as PMI
DOGBERT SAID: shouldn't really render as DOGBERT SAI

either...
When the dumb XenForo parser renders any emoticon markdown shorthand
even when not bracketed by spaces or similar characters like EOL, it's the parser itself that's IMHO broken. The Ars forums are often technical, with people quoting scientific papers and other stuff that might have all sorts of : / \ ! characters inside words that don't really belong in code or icode brackets...