the key problem is that what they delivered is not needed by market.
I doubt that. Until now the blockchain's USP, as I wrote, wasn't delivered (not because it's a scam but because they clarified that at their roadmap, from the beginning). So
currently the Ardor child-chains do not offer significant advantages to other "token" solutions.
But once child chain pruning is delivered (testnet is planned still for late 2019 and mainnet for 2020), that would provide a crucial "maintainance cost" advantage to other blockchains, as initial blockchain download and full node RAM/disk space requirements would be much lower than for coins which do not offer that feature (the only exception is actually Cryptonite, an even much more smaller altcoin which has a similar system but without child chains).
they need to accept advices from community to delivered a much suitable product.
Yep, that's just what I'm criticizing too: the decision-making is too centralized. They however accept some advices (a comment
I made at Stackexchange probably inspired minimal-trust lightweight contracts, as they implemented a technique Lior had described in his answer) and perhaps even
pull requests as they write on their Bitbucket repo.
But generally speaking, I think the coin has not a problem related to missing features, but more of a communication problem. This thread, for example, is practically deserted, the
Ardor forum has low activity, and there is confusion between the NXT and ARDOR brands (while Ardor isn't a particularly bad name, I would have preferred it to be called simply "NXT 2.0" as originally intended, but they cited intellectual property problems because of other NXT-named brands). And many ask themselves what the IGNIS token is meant for - the answer is actually that its utility depends also on the pruning feature as it would enable a lightweight payment sub-chain.