Basically callbre means that we call replicaset with two additional parameters:
    1) prefer_replica = true. We can use read from replicas because
        transactional consistency is out of Sbroad scope. Basically we
        do not care about data consistency at all.
    2) balance = true. This parameters enables internal vshard load
        balancing between replicaset nodes in round-robin fashion,
        with a preference for replicas if prefer_replica=true is also set

We have fixed some bugs about NULL and made equality propagation
use the infrastructure common with the other transformations.

Also, extended and refactored the tests.

Merge 2 execute query functions in one. Response waiting timeout was extracted to cluster config.

Was made determining uniq set of replicaset by set of buckets that avoid send the same queries to the same replicas. Bucket exec query function was fixed to async mode.

Closes an error in the backend AST reordering logic

Now we can eliminate buckets with multi-column (i.e. composite)
keys in equalities. Previous implementation was incorrect, so
we have to make an executor refactoring as well.

Add support join query to linker, now executor can execute join query with distribution conflict (plan has motion nodes). Extend text in some error

Join node deduces the distribution of its output tuple based on
the output tuples of its left and right children. The logic of
deduction was broken: if we swap the children the distribution
changed. So, current commit rewrites join distribution deduction.

Also we refactor some parts of the set_distribution() function.
Its code style is still bad, but now it is not so horrible as
before.