Add optimizations for primary key queries

[DCNick3]

TL;DR

After #4833 lands, there will be a degradation in performance of primary-key queries compared to before (linear scan is being used instead of looking up a map). An special case should be created for the primary-key queries, as outlined in https://github.com/hyperledger/iroha/pull/4833#discussion_r1679031539:

let iter: Box<dyn Iterator<Item = Domain>> = match filter {
    CompoundPredicate::Atom(DomainPredicateBox::Id(DomainIdPredicateBox::Equals(
        domain_id,
    ))) => Box::new(Some(state_ro
        .world()
        .domain(&domain_id)?
        .clone())
        .into_iter()),
    filter => Box::new(
        state_ro
            .world()
            .domains_iter()
            .filter(move |&v| filter.applies(v))
            .cloned(),
    ),
};

Ok(iter)