Oh boy, difficult subject. But you're totally right, it needs to be discussed!
I've actually spent quite some time pondering about this recently, and here are some of my thoughts so far. That's just a temporary snapshot of stuff floating in my head, to be honest. 😉
First of all, I don't think the outlook of AIR is as grim as some people believe, now that the Flash Player EOL has been announced. That Flash itself had no longer a place in modern browsers has been obvious for several years now — even before the EOL announcement, only few developers started web-based projects with Flash recently. Adobe just finalised what was happening anyway.
AIR is still excellent in the job that it's supposed to be doing: comfortable cross-platform development. And Adobe is still doing a lot of work on the platform (just this week, AIR 27 was released), even though we won't be seeing the *huge* changes we developers are craving. That's just a fact: the language, as you said, is more or less frozen; the API is not.
What's more serious, in my opinion, is that many developers have lost faith in it; new developers won't learn AS3, and clients will often not accept AS3-based solutions because, heck, Flash is dead (and what's that AIR thingy, you said)?
If Starling, of all frameworks, switches to Haxe, I feel like I'd be giving it the final blow.
(Yeah, I know people can still use the SWC with AIR. But that doesn't change the message.)
That doesn't mean that it wouldn't make sense to have Starling for Haxe — I'm extremely happy to see that there is already a port, for all the good things that can come from it. I'm very proud of the API, and I love to see as many people as possible work with it.
Whether it's my "official" Starling or another developer's port: all people would need is a good and extensive document that explains how to migrate over to Haxe, the main differences, the choices when it comes to IDEs and compilation, etc. That's what I'd like about this route.
On the other hand, though, there is already OpenFL — an API that is also based on Flash and widely used by developers. What, and that's honestly a question I'm interested in, would I bring to the table by moving Starling to Haxe? Would people outside the Starling community (!) start new projects with Starling.hx? Or would they rather use tried-and-tested OpenFL (or Unity, ReactNative, ...)?
If the reason for switching to Haxe is just to make it easier for people to move content from AS3 to Haxe, then I fear that's a too small gain to get myself excited enough about it to pull it off; and to finance it, for that matter — Adobe's financial support is really what made Starling (and Feathers) possible over such a long time.
You know, when I started working on Sparrow and Starling, I felt that I'd be providing something really new and that my work actually made a difference. That's not how I'm feeling about such a port, and that's probably why it doesn't immediately "click" in my head, even though I really tried to make myself like it.
Those are my current thoughts on this topic — as I said, nothing is carved in stone, but that's how I feel about it right now, today, in this minute. 😉