fix(copilot): remove is_long_running hack from agent generation tools

Remove the `is_long_running = True` override from create_agent,
edit_agent, and customize_agent tools. Now that CoPilot runs in the
executor service (which already handles background execution), the
async delegation pattern is unnecessary.

This fixes the issue where agent generation completion messages
never appeared in chat because the code was exiting early expecting
an external Redis Stream completion that never came.

The tools now execute synchronously in the CoPilot executor and
stream completion messages back to chat immediately.

Fixes: Agent generation completion not showing in chat

autogpt_platform/backend/backend/copilot/tools/create_agent.py

@@ -47,10 +47,6 @@ class CreateAgentTool(BaseTool):
     def requires_auth(self) -> bool:
         return True
 
-    @property
-    def is_long_running(self) -> bool:
-        return True
-
     @property
     def parameters(self) -> dict[str, Any]:
         return {

autogpt_platform/backend/backend/copilot/tools/customize_agent.py

@@ -46,10 +46,6 @@ class CustomizeAgentTool(BaseTool):
     def requires_auth(self) -> bool:
         return True
 
-    @property
-    def is_long_running(self) -> bool:
-        return True
-
     @property
     def parameters(self) -> dict[str, Any]:
         return {

autogpt_platform/backend/backend/copilot/tools/edit_agent.py

@@ -45,10 +45,6 @@ class EditAgentTool(BaseTool):
     def requires_auth(self) -> bool:
         return True
 
-    @property
-    def is_long_running(self) -> bool:
-        return True
-
     @property
     def parameters(self) -> dict[str, Any]:
         return {