Add call_activity_with_retry to DurableOrchestrationContext class

💡 **Feature description**
The `DurableOrchestrationContext` class has both a `call_activity_with_retry` method and a `call_sub_orchestrator_with_retry` method. Often times the call_http method fails for unknown and external reasons. Adding a `call_http` method that includes an option to retry a specified number of times would go a long way to improve the reliability of durable orchestrations. Adding an exponential backoff would be a nice plus.

💭 **Describe alternatives you've considered**
The obvious alternative is to wrap your API call into an activity function and use the `call_activity_with_retry` method. Creating new activity functions for each API call tends to bloat projects though. The v2 programming model may address this concern though - I'm still stuck on the v1 programming model for now.

Commencez par examiner la classe DurableOrchestrationContext et ses méthodes existantes call_activity_with_retry et call_sub_orchestrator_with_retry. Comparez leur comportement avec call_http et définissez ce que doit signifier un nombre de tentatives spécifié ; la finalisation doit permettre de réessayer les appels HTTP, avec un backoff exponentiel comme amélioration facultative.

Rédigé par le modèle d'indexation à partir du texte de l'issue.