Remove use of timestamps in failover code

Using timestamps to detect whether MaxScale was active or passive can
cause problems if multiple events happen at the same time. This can be
avoided by separating events into actively observed and passively observed
events. This clarifies the logic by removing the ambiguity of timestamps.

As the monitoring threads are separate from the worker threads, it is
prudent to use atomic operations to modify and read the state of the
MaxScale. This will impose an happens-before relation between MaxScale
being set into passive mode and events being classified as being passively
observed.

include/maxscale/monitor.h

@@ -205,8 +205,7 @@ struct mxs_monitor
     bool active; /**< True if monitor is active */
     time_t journal_max_age; /**< Maximum age of journal file */
     uint32_t script_timeout; /**< Timeout in seconds for the monitor scripts */
-    int64_t last_master_up; /**< Time when the last master_up event was triggered */
-    int64_t last_master_down; /**< Time when the last master_down event was triggered */
+    bool master_has_failed; /**< Set to true when the latest event is a master_down event */
     struct mxs_monitor *next;     /**< Next monitor in the linked list */
 };