The HealthCheck infrastructure (see https://github.com/mongodb/mongo/blob/master/src/mongo/db/process_health/fault_manager_config.h) can be configured to crash mongos when it is enabled and a stall or lack-of-health is detected. 

When a router is embedded with a shard, we should be more conservative and likely never crash the server. 

We should ensure that the ProcessHealth system cannot be configured to crash the system when it is running on an embedded router, or cannot be enabled for embedded routers at all.