It's not the first time someone has started building something on top of LN. There was a previous effort, its main target was to do with anonymously renting compute. It didn't take off.

I think that the layer cake has another layer they are missing, which relates to programmable relaying services.

In general, I don't think that public hosting of "smart contracts" is very smart at all, in fact, it's not gonna be practical until ZKP homomorphic computation is cheap enough. That's a really high target compared to current computation hardware capabilities.

I think Nostr and our project Indra are going to be hybridised in the near future (we are planning to make indra by default a nostr relay, based on fiatjaf's original Golang based relay). A programmable relay network also can provide solutions for asynchronous, trustless lightning payments as well, which is critical because the liveness requirement of LN is so high.

Circling back to the question, these capabilities are the missing ingredient for a system like RGB.

Really, what we are witnessing is that the base concept of LN state channels have a whole field of variations that enable many other things, but the first thing, after payment, is relaying messages.