Add tests for issues with too many queue wakeups

[fwsmit]

Some tests should be added that can test for any regresions in the following issues:

• High CPU usage for updating notification timestamps #1102
• High cpu usage when show_age_threshold = -1. #1163
• Regression by mixup of n->timestamp and n->start #1166

[bynect]

How would we go about testing this? Do we need to query the current cpu usage?

[fwsmit]

No, you can just query the function that says how long Dunst should sleep. If the sleep time is to short, the cpu usage is high