The hoverover/help explains this is 1 cpu. So 100% of 1 cpu if this is a 2 cpu system you would consider this 50%.
Looking at your grafana that system is often oversubscribed. Load avg is exceeding your cores. Just after 17:05 the load avg really climbs. The consistent thigh load would explain a slow-down. The upramp at the end I have experienced as a system lock up.
The mongo monitoring looks nominal to me, making me think there is another factor on the host creating this load. Another container or process perhaps.