Piggybacking onto curtisi's comment here, if you are trying to think of this in the same way you would do alerts in Orion you will become very frustrated because it works on entirely different principles. LEM is event driven, so you can't set up behaviors like "alert me until this is acknowledged" because LEM doesn't keep track of things like the state of an alert (you can kind of do that with state variables but that's maybe a bit of a tangent.) The first method described in the video would not work for events that don't generate their own sort of "clear" event such as the calc process stopping in his example. For the disk space event windows automatically rechecks every 5 mins (this is a windows thing, no way to control it with lem) and any time you are above the threshold it generates a fresh event. You can use the trick with the re-infer to stop getting spammed but that method can be tricky because, as he describes in the video, having a long re-infer period, or having it in play across lots of rules/devices can be really hard on your appliance resources.
You want to get used to thinking of LEM in terms of when this event is generated do X. I would also say if you have NPM/SAM then you may as well turn off the volume space alert in LEM and rely on Orion for those because LEM is not nearly as efficient for that kind of alerting/monitoring.