Generic methodology to simulate crash for all tests

[hkadayam]

We need a generic mechanism to simulate crash and subsequent recovery for all unit tests to run. Without this, we are relying on actual crashing and then restart with some script. The pain point is debugging as well as automation suffers.

One idea is to simply return success on all write operations in iomgr after the simulation point and then do clean shutdown and restart. That should simulate the crash scenario.