On the contrary, #5's actually the most likely one on there, since it's the only one Riot's mentioned they're "actively discussing." Riot's explicitly said permanent Featured Modes aren't happening, besides maybe a queue rotating between various Featured Modes; visual updates would take a massive amount of work, but could happen when Riot finally becomes embarrassed at how outdated TT and CS are after a few more SR updates. There's no way LCS-TT or LCS-Dom are happening before visual updates and even then would require Riot to train/hire a bunch of people who actually know those maps. AI difficulty's probably low-priority on a to-do list somewhere. In order of decreasing likelihood, it's 5 > 4 >> 2 >> 1 >> 3.
In general though, Riot's probably not going to support alternate maps much, if at all. They went all-in on SR years ago. The main problem is the same one it's been for ages: Riot doesn't support alternate maps, so players don't support alternate maps, so Riot doesn't support alternate maps. It's a difficult issue to address precisely because of how long Riot's let it persist. Riot would need to consistently demonstrate a desire to support alternate maps in order to win back player trust. However, this would take a long time, and Riot doesn't want to throw that many resources into something that's not SR.
Dynamic queue's raised an interesting question though. In the past, one of the reasons Riot used to deny a ranked Dominion queue was that the vast majority of League's playerbase wasn't interested in ranked, and so a ranked queue wouldn't significantly increase Dom's playerbase. Coincidentally, one of the reasons Riot used to introduce Dynamic queue was that the majority of League playerbase is interested in ranked, but not in the current ranked system. I'm curious how the new ideas behind Dynamic queue would work on the Crystal Scar.