This task  is about 'get' requests distribution between primary and
backup nodes in the replicated cache if 'readFromBackup' flag is
I've prepared a solution  suggested by Alexei Scherbakov in Jira
comments. It passed prereviews by Alexei and Nikolay Izhikov.
TeamCity tests look similar with the master branch.
Could someone of core module maintainers do the final review ?
Best Regards, Vyacheslav D.