Problem description: for some reason, when filtering events for which CEP operations are executed, the event that got evicted due to exceeded capasity in the window is considered as the last event in the time window.
We should consider the last event that has successfully fitted into the time window as the last event in the window.

Steps to reproduce:

1. Import the following host: host_with_3_triggers.yaml
2. Create CEP rule with the following configuration:
   • Conditions: Tag name equals test
   • Time window: Simple (multiple)
   • Duration: 1m
   • Capasity: Limited => 2
   • Operations:
     • First operation:
       • Execute when : Event evicted
       • Operation: Close
       • Applies to: All events
       • Filter: Last event in time window
     • Second operation:
       • Execute when: Event evicted
       • Operation: Clone last
       • Applies to: All events
       • Filter: Last event in time window
3. Send value "3" to the trapper item on the imported host:

   ./zabbix_sender -z 127.0.0.1 -s test -k trap -o 3
   

4. Check Monitoring => Problems view: two problems were opened, so window is now fully occupied
5. Send value "4" to the trapper item on the imported host:

   ./zabbix_sender -z 127.0.0.1 -s test -k trap -o 3
   

6. Check Monitoring => Problems view

Result: the evicted event "Trapper trigger 3" got closed and cloned:

Expected: the last event that fits into the time windows should be Closed and Cloned ("Trapper trigger 2")