Extra calls to check_completed_func after SetupTask is run

It appears the check_completed_func is called again after carthage has just run a setup task. This seems wrong.

Output detailing multiple calls to check_completed_func from running layout:
```
PYTHONPATH=/home/srak/project/locallayout/carthage: /home/srak/project/locallayout/carthage/bin/carthage-runner --no-pull-plugins --no-tmux . --command-verbose --tasks-verbose --generate
Carthage console. Type Python expressions or use !help for Carthage commands.
>>> !deploy localhost -y
: process started
Task Touch a file source modified at 2025-07-11T04:57:32.178738 more recently than last run of 1969-12-31T19:00:00.
Running cust task for
: process started
Task Touch a file source modified at 2025-07-11T04:57:32.178738 more recently than last run of 1969-12-31T19:00:00.
Running Touch a file task for
: process started
Finished running Touch a file task for from 05:00:44.766125 to 05:00:44.771827 (0:00:00.005702)
Finished running cust task for from 05:00:44.753085 to 05:00:44.774049 (0:00:00.020964)
: process started
Task Touch a file for determined complete by check_completed_func(); no timestamp provided
Running cust task for
: process started
Task Touch a file for determined complete by check_completed_func(); no timestamp provided
Finished running cust task for from 05:00:44.785275 to 05:00:44.795789 (0:00:00.010514)

# Deployment Report

Ran deploy on the following objects successfully:

*

Successful deploy successes:1
>>>
```

MVE layout to reproduce:
```
import sh

from carthage import *
from carthage.modeling import *
from carthage.network import V4Config
from carthage.local import LocalMachine

class layout(CarthageLayout):
layout_name = "testlayout"

add_provider(machine_implementation_key, dependency_quote(LocalMachine))

class localhost(MachineModel):
name = "localhost"

class cust(MachineCustomization):
@setup_task("Touch a file")
async def touch_file(self):
await self.run_command("touch", "/tmp/foo")
@touch_file.check_completed()
async def touch_file(self):
try:
await self.run_command("ls", "/tmp/foo")
return True
except sh.ErrorReturnCode_2:
return False

```

Start with the provided MVE layout and reproduce the deploy using the setup_task and check_completed() definitions on MachineCustomization. Trace the calls reported around SetupTask completion and check_completed_func, then verify that the completed task is not checked again after setup runs.

Written by the indexing model from the issue text.