[PECO-1532] Ignore the excess records in query results

[kravets-levko]

PECO-1532

Note for reviewers: This PR contains some refactoring needed to implement the fix, so probably it's easier to review commit by commit

When client library executes query and wants an Arrow-based or Cloudfetch results - server will return records as Arrow batches. Batch size may vary, server makes the decision on that depending on count of records, record size, etc. But usually all batches will have the same size, with the only exception - the last batch, which usually contains less records. And there are two possibilities:

• server may make the last batch smaller and containing only the remaining records;
• server may actually fetch more record than needed to make the last batch of the same size as others. But it also sets a rowCount field which defines how may "valid" records are in the batch. Client should use only that records and discard the remaining ones.

(I guess that different workspaces may be configured differently and will behave either as described in scenario 1 or 2)

Nodejs connector doesn’t use value from rowCount and therefore returns that extra records to user. This behavior is wrong, and this PR fixes it.

[benc-db]

Looks reasonable to me, but I'm neither a node nor thrift semantics expert.