This is a refreshingly honest analysis from someone that has a lot of skin in the game. I completely agree on both points. First, the functional paradigm is important and a step forward for most problem domains. Second, none of the current crop of functional languages is going to truly break into the mainstream.
I think it's more likely that we'll see more and more functional techniques trickle down into mainstream languages than it is that we'll see a pure functional language adopted as-is. C++ just picked up lambdas, after all. I do regret somewhat my dalliance with FP over the last decade though. If I'd invested that time in mastering DSP or graphics programming instead I'd be better off overall.