Tasks with all_done trigger rule execute when they should skip

### Apache Airflow version Other Airflow 2 version (please specify below) ### If "Other Airflow 2 version" selected, which one? 2.10.5 ### What happened? I have a task that filters a list based on values of a second list. As such it depends on two upstream tasks, each producing one of the lists. The second upstream task is mapped dynamically. With the default trigger rule `all_success` the task in question will skip if any of the mapped tasks skips, which feels counter-intuitive. With the `all_done` trigger rule the task behaves as expected, waiting for all mapped tasks to finish and executing with the resulting mapped list of outputs. However, if _all_ of the mapped tasks skip, which is a valid result in this context, the tasks still tries to execute. This might be related to #51320. ### What you think should happen instead? The task receives the original list and instead of the second list with filter values it receives `None`. I think this happens because the mapped return values of skipped tasks (`None`) are reduced to `None`. In my opinion it would be more appropriate to reduce them to `[]`. Apart from that, the task should just be skipped. ### How to reproduce ```python from random import random from typing import Any, List import pendulum from airflow.decorators import dag, task from airflow.exceptions import AirflowSkipException from airflow.models.param import Param @dag( "playground", "Try things", schedule=None, start_date=pendulum.now(), params={ "p_abort": Param( type="number", title="p_abort", minimum=0.0, maximum=1.0, description="Probability to abort mapped tasks", ) }, ) def playground(): @task.python def produce_list() -> List[int]: return list(range(25)) @task.python def maybe_abort(value: int, **kwargs) -> int: if random() > kwargs["params"]["p_abort"]: return value raise AirflowSkipException(…