why - without a test function, I find I'm often breaking the entire application when I create new health checks due to some minute syntactical error like a miss spelling of the path or expecting a 301 instead of a 302, or forgetting to include a required header. The test function would ensure you can confirm the health check works before applying it only to find out if fails 100% of the time and causing a 100% application outage because all POP paths are unhealthy and the LB is not routing any traffic to the origin.

who benefits - first and foremost the application user benefits by not having a 5 to 15 minute down time of the app while the XC engineer scrambles to either remove the health check or fix it to get the app back up. second the XC engineer gains peace of mind knowing their traffic should continue to flow to the origin because the health check was confirmed working before being applied.

how it should work - similar to how the test button works on GLR (global log receivers) it should send a transaction to the origin as configured and display the results and indicate if the result matched the configured expected outcome or not. if not, it should warn you if you try to save or simply not allow the save. It should allow the engineer to select which POP path to attempt the test from or simply attempt it from all POP paths and then let the engineer decide to save as is or make changes before proceeding. Ideally, the save option would not be available until a test has been completed.