The crazy thing is that flash still has amazing penetration over Unity on the web. Now, I can't seem to find an accurate statistic regarding the plugin penetration of Unity recently, but my guess is that it's still significantly lower than flash's penetration. This is what I see as the major thing flash still has going for it. Flash still has even more penetration than HTML5. If you are a major Facebook game developer, for instance -- you care about this. Each percentage point you lose = millions of players.
Unity claims that 1/3rd of Facebook gamers have the Unity player installed. So 33% of Facebook 'gamers' (What is it defining as a 'gamer' anyway?) have it installed. Compare that to 99% for flash. Facebook is a great demographic, were talking about the most casual game player. The people who you might actually want to play your game perhaps ?
We're all nerds here so we're all focused on the technical aspects of the platforms, but if you look at it from a business perspective Flash still makes a lot of sense if you're talking about web-only.
BTW I'm guessing this was not totally trivial on Unity's part to maintain this flash compilation process. I think they were using FlasCC, and I'm not sure how well Adobe was supporting this. I bet Unity's engineers were pulling their hair out to try to port every Unity feature over to it. With their own plugin gaining momentum they did the right thing to divert the resources to somewhere else.