Update current User query parsing code to accomodate multiple different asset types

[marcel-vesely-kinit]

The current implementation of user query parsing using an LLM is limited to only dataset asset type, hence the need to modify it to be more generic.

More precisely, you should be able to plug the asset type and its corresponding schema an LLM should adhere to as arguments once you invoke an LLM.