Nope, the severity is hard coded into the alert. The other common spot where it would be nice to have things be a bit more flexible is the alert trigger times, x must be true for y minutes. Unfortunately both of them are static and cannot be changed without creating a new alert.
I have to admit, I have been subscribing to the school of "there are no warnings" when it comes to my alert messages. Anything that falls into a traditional warning range gets pushed off to a once a day report that shows which devices breached their warning thresholds, and for how long. Actual alerts only go out when there is an actionable issue that requires immediate human intervention for my environments.
Thanks for the reply.
As far as poling intervals check out the new feature with 2019.2 for variable polling. I have a feature request to get exactly what you mentioned in the laerts "Default Thresholds for Consecutive Polls in Sustained Thresholds "
I agree that there are no warning. I am using severity to map to my CMDB/ticketing tool (service-now) and that controls the notification method (call out, vs generating ticket).