fix: use seconds value for retry_after

[iprak]

The value assigned to UpdateFailed exception (in #161) is supposed to be seconds and not timedelta() which was resulting in following exception.

2026-04-19 10:04:52.913 ERROR (MainThread) [homeassistant] Error doing job: Task exception was never retrieved (task: None)
Traceback (most recent call last):
  File "/usr/src/homeassistant/homeassistant/helpers/update_coordinator.py", line 296, in _handle_refresh_interval
    await self._async_refresh(log_failures=True, scheduled=True)
  File "/usr/src/homeassistant/homeassistant/helpers/update_coordinator.py", line 563, in _async_refresh
    self._schedule_refresh()
    ~~~~~~~~~~~~~~~~~~~~~~^^
  File "/usr/src/homeassistant/homeassistant/helpers/update_coordinator.py", line 270, in _schedule_refresh
    next_refresh = int(loop.time()) + self._microsecond + update_interval
                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~
TypeError: unsupported operand type(s) for +: 'float' and 'datetime.timedelta'

To test define this static field on class WinixDriver: load_test_counter = 0 and this block to the beginning of get_state:

        WinixDriver.load_test_counter += 1

        if WinixDriver.load_test_counter == 2:
            raise WinixTransientError(f"{self.device_id} simulated transient error for testing")

[iprak]

@mfncl9991 FYI, it seems we didn't test #161 well enough.

[mfncl9991]

@mfncl9991 FYI, it seems we didn't test #161 well enough.

Thanks for the heads up. Noted. Anything specific to generate the error, or just simply running the integration with the changed files in the branch?