Thank you for your patience here.
I did a lab and observed the same health probe behavior as you have observed above and it is an expected behavior of Azure Traffic Manager Health probe monitoring.
Since the App Service is enabled with zone redundancy which has 3 instances alive. Azure Traffic Manager will probe the all the three instances of the App Service.
I was able to validate the observation above by collecting a packet trace at my Web App the packet trace will generate three files for each instance and you can validate the health probe packet as user agent will be Azure Traffic Manager Endpoint Monitor (Reference Image below)
Another aspect of this is that Traffic manager deploys multiple probes from multiple locations for each endpoint. Multiple probes increase resiliency for endpoint monitoring. Traffic manager aggregates the average health of the probes rather than rely on a singular probe instance. The redundancy of the probing system is by design.
This is currently documented here
Hope this helps! Please let me know if you have any questions. Thank you!
Please "Accept the answer" if the information helped you. This will help us and others in the community as well.