Using distributed tables with Materialized views

[jehuty0shift]

I hereby agree to the terms of the CLA available at: https://yandex.ru/legal/cla/?lang=en

This commit allow to use materialized view to ingest data to distributed tables. The pipeline ingestion is the following:
Kafka consumer node => Materialized View => Distributed Table.
Using Distributed Table instead of local tables has two main advantages:

1. the underlying database can scale independantly of the kafka consumers nodes. When using only local tables, it is more complicated to achieve this.
2. the data stays evenly/distributed in case of a kafka consumer node failure. this is not the case when you use only local tables.
   These two points allow a highly available and balanced ingestion pipeline.

Unlike previous attempts to allow this feature, the change in code is minimal (and fix a potential NULL pointer dereferencing) since only StorageDistributed.cpp is modified, however i wrote an integration test (heavily inspired by another one) to ensure that this functionality does not break into the future.

[alexey-milovidov]

Thank you!