That's been the discussion so far, the thing, if I'm not mistaken (BIG IF as I'm no coder) is that recursive covenants are only recursive within themselves, so the worst that can happen is that they eat up the memory of a node trying to validate the whole TX because of the recursive execution of the script, but there's been a lot of FUD around recursive covenants being able to propagate themselves as if Bitcoin was an account model chain and not an UTXO model where only the receiver writes the spending conditions.

When it comes to CTV+CSFS lack of demand's argument, I believe Steven Roose makes the best counter-argument: