🐛 Describe the bug
Azure Durable Functions Python SDK fails to serialize custom objects that implement a to_json() method when returning from orchestrator functions. The error "Object of type [CustomClass] is not JSON serializable" is raised.

🤔 Expected behavior
Custom objects implementing a to_json() method should be automatically serialized when returned from orchestrator functions.

☕ Steps to reproduce
Basic orchestration with custom return types.

1. Create a custom class with to_json() and from_json() methods:

from dataclasses import dataclass

import azure.durable_functions as df
import azure.functions as func

app = df.DFApp(http_auth_level=func.AuthLevel.ANONYMOUS)

@dataclass
class CustomResult():
  message: str
  success: bool = True

  def to_json(self):
      return {"message": self.message, "success": self.success}

  @classmethod
  def from_json(cls, data):
      return cls(message=data["message"], success=data["success"])

@app.orchestration_trigger(context_name="context")
def orchestrator_function(context: df.DurableOrchestrationContext):
  return CustomResult(message="Hello from orchestrator!", success=True)

@app.route(route="test", methods=["GET"])
@app.durable_client_input(client_name="client")
async def http_start_conversion(req: func.HttpRequest, client: df.DurableOrchestrationClient) -> func.HttpResponse:
  instance_id = await client.start_new(
      orchestration_function_name="orchestrator_function"
  )

  return client.create_check_status_response(req, instance_id)

2. Execute the orchestration - it will fail with "Object of type CustomResult is not JSON serializable"

Environment: Running locally with
Azure Functions Core Tools Version: 4.1.0
azure-functions-durable: 1.3.2

Root cause: In TaskOrchestrationExecutor.get_orchestrator_state_str(), the code calls dumps without a default argument:

But should call:
json.dumps(self._data, default=_serialize_custom_object) as in other models or entities.

⚡If deployed to Azure
Didn't test yet. Since the bug is in the Python SDK code itself, I expect the same issue to occur.