Great points.
But let's not forget that those second-layer tokens have to exist before we even begin the discussion of removing the reward pool. And before that the technology to build the tokens has to be developed. HiveEngine is not an acceptable option; it's just a single centralized database. I guess that didn't stop anyone; counting those chickens before they hatch.
I tend to agree. I was not a fan of SteemEngine when it first came out, but I consider it now a bit like a "proof of concept". Yes it's centralised but it's allowing some innovations to prove that they can work. Hopefully one day we'll get SMTs, the worthwhile tokens will be able to migrate over and get implemented properly.