[MEX-693] Performance optimisations on user queries

[mad2sm0key]

Reasoning

1. In the current implementation, the userNfts query uses a pending executor to ensure a sinlge API call is made for fetching the users MetaESDT tokens. If the response time from the API is fast, the pending exector will still make multiple identical requests to the API (it does not guarantee a single request)

2. Token identifiers returned by SC views are data points that don't change often. The current implementation uses a cache TTL of 10 minutes, which means (usually) API calls for fetching token IDs from contracts happen often

Proposed Changes

• perform a single API request for fetching all of the users MetaESDT tokens in the root level userNfts query and store the result in the UserNftsModel. The resolve field methods are able to use this without performing extra requests
• add special CacheTtlInfo property for token IDs with a TTL of 1 day
• add caching to getNftByTokenIdentifier API call

How to test

• N/A