1. sleep(0.1) -> sleep(0.01)
2. _server -> serv, _replica -> _repl.

1. Not waiting for flush_all
2. Delete redundant time.sleep() and wait()'s
3. Fix for cfg files

2. Delete raw sleeps from replication-memcached test
3. Add check, that replica expires even when it becomes master

2. Add expiration tests
3. Move tnt-memcache configs to cfg folder