Operation tests hurt from fake time coupled with log file rotation
- Operation tests launch plenty of hived instances. Each of them spawns a separate logging thread.
- It's been discovered that each of the logging threads takes up to 100% processing core time.
- Most probable reason is that fc scheduler works poorly with fake time (used in these tests).
- This MR turns off log file rotation for sped up nodes, eliminating rotation task scheduling and thus the excessive load of logging threads.