Make ConnectionPool::SpecificPool::updateEventTimer topology reconfiguration aware

BF-20554 surfaced a problem where ConnectionPool::SpecificPool::updateEventTimer might incorrectly trigger deferredStateUpdateFunc while a topology reconfiguration is happening in the shard leading to client's DoS.

The logic might have to consider when a topology reconfiguration happens and act appropriately.

AC: After exceeding kHostRetryTimeout (currently hardcoded to 1 second), log how long it takes for getting a connection within the time limit before raising ErrorCodes::NetworkInterfaceExceededTimeLimit