Great article, Phil. I just worry about such improvements by MS especially if they resort to using .NET functionality behind the scenes.
For example, look at what they did with the FORMAT function in SQL Server. It's something that a whole lot of people have wanted for ages. When they finally do get it, it works correctly but a huge cost... it's an average of 44 times slower than CONVERT. If that's the way that its backbone works in .NET, then .NET also has a problem. It seems like they may have used the awful "regex-replace" technology instead of something much simpler and much faster.
BWAAA-HAAA!!! Reminds me of an old saying with a different twist on it... "Just because you CAN do something in .NET, doesn't mean you should! 😛
And, I think you're asking a lot from a company that can't even release a string splitter function (STRING_SPLIT) without realizing what the necessary functionality should have been (missing ordinals of the split-out elements).
That being said and if they do take on the problem(s) you speak of in your good aritcle, I hope they write separate functionality rather than try to incorporate it into LIKE, CHARINDEX, and PATINDEX... there's just too great a chance that it'll come out with performance penalties like FORMAT or half-baked like STRING_SPLIT and I don't want stuff that's currently working well to suddenly become an incurable problem.
Heh... anyone remember when they first came out with MERGE? Yeah... that was a "joy".
"Change is inevitable... change for the better is not".