bug fix: DateTimeSensor can't render jinja template if use native obj

[nailo2c]

Closes: #38336

🐞 Problem

This is quite an interesting case.

• With render_template_as_native_obj=True, the Jinja expression is stored as a raw string when DateTimeSensorAsync is instantiated
  {{ (data_interval_end + macros.timedelta(minutes=1)) }} → falls through line 91 and is kept as‑is.
• At run‑time the template is rendered to a pendulum.DateTime object.
  When execute() calls

  moment = timezone.parse(self.target_time)

  timezone.parse expects a string and raises TypeError.

✅ Fix

• Add a private helper _moment() that
  • returns self.target_time directly if it is already a datetime / pendulum.DateTime;
  • otherwise parses the string with timezone.parse.
• Replace all direct uses of timezone.parse(self.target_time) with _moment().
• No behaviour change for DAGs that keep render_template_as_native_obj=False.

📝Example DAG (runs successfully after this patch)

from datetime import datetime
from airflow import DAG
from airflow.providers.standard.sensors.date_time import DateTimeSensorAsync

with DAG(
    dag_id="test_date_time_sensor",
    start_date=datetime(2025, 1, 1),
    schedule=None,
    render_template_as_native_obj=True,
):
    DateTimeSensorAsync(
        task_id="wait",
        target_time="{{ (data_interval_end + macros.timedelta(minutes=1)) }}",
    )

[Lee-W]

Left a few nits. In general, it looks good. I won't mind if someone wants to merge it directly

[nailo2c]

Hi @Lee-W, if everything still looks good to you, could you please merge it at your convenience? That would close #38336.
Thanks a lot! 🙏

[dwreeves]

For whatever it's worth, looks good to me. Thanks for addressing the issue! I can't wait to remove all those | string's from my code 🙂

[nailo2c]

I'm so glad I could help with this issue 😄