And don't forget that many use cases that have gone Ethereum due to market effect could have perfectly used Nxt.

Associating smart contracts with Ethereum and not with Nxt/Ardor, in this sense, is a fallacy. A smart contract allows for self-enforcing rules in a blockchain, and the Nxt architecture provides modules that allow for self-regulating contracts (specially the whole phased transactions API) and other modules that work as a template for frequent uses in Ethereum, like asset/currency issuance and trade.

Turing completeness in Ethereum surely does allow in theory for virtually limitless custom tools, that is undisputed. But the idea behind Nxt/Ardor is modularity, which can simplify implementation of a lot of common cases, even if theoretically limited in scope so far. Enter the scalability provided with Ardor, which in the case of Ethereum (plasma, sharding...) still has a long way to go, and it is quite clear that, if we all put our fanboy hats aside, Ardor can and will be a competing technology.