feat(models): expose partition_key and partition_date on TaskInstance

[bujjibabukatta]

Add Asset partition attributes to TaskInstance
Fixes #68079, related to #68075.
What changed:
airflow-core/src/airflow/models/taskinstance.py — adds a partition_date property that parses partition_key (already proxied from dag_run) into a timezone-aware datetime, returning None for non-partitioned runs or unparseable keys.
airflow-core/src/airflow/utils/helpers.py — fixes log_filename_template_renderer in two ways:

Jinja branch now injects ts_nodash, ts, ts_nodash_with_tz, partition_key, and partition_date into the render context instead of returning a bare Template.render with no context.
f-string branch no longer crashes when logical_date is None (falls back to partition_date, then empty string).

Why: Asset-triggered DAGs using PartitionedAssetTimetable set logical_date=None, causing the scheduler to crash on ti.logical_date.isoformat() during log filename rendering. Users also had no way to reference partition attributes in templates at all.

[jroachgolf84]

I don't think that this is a good way to go about this. partition_date is a field that's already available as part of the DagRun model. We shouldn't need to impute it.

I had been playing around with this locally, and kinda came to the conclusion that I don't think it makes sense to add it. partition_key should always provide the info that is needed, we don't really need to add the partition_date.

For what it's worth, I used this DAG to test:

from airflow.sdk import (
    Asset,
    DAG,
    task,
    CronPartitionTimetable,
    PartitionedAssetTimetable,
)
from datetime import datetime

fix_issue_68079 = Asset(name="fix_issue_68079")

with DAG(
    dag_id="fix_issue_68079__upstream",
    start_date=datetime(2026, 1, 1),
    schedule=CronPartitionTimetable("15 * * * *", timezone="UTC"),
) as upstream_dag:

    @task(outlets=[fix_issue_68079])
    def transient():
        pass

    transient()


with DAG(
    dag_id="fix_issue_68079__downstream",
    start_date=datetime(2026, 1, 1),
    schedule=PartitionedAssetTimetable(assets=fix_issue_68079),
) as downstream_dag:

    @task
    def output_context(**context):
        print(f"context['ti']: {context['ti']}")
        print(f"context['dag_run']: {context['dag_run']}")
        print(f"context['partition_date']: {context['partition_date']}")
        print(f"context['partition_key']: {context['partition_key']}")

    output_context()

[jroachgolf84]

@bujjibabukatta - I closed the issue as "not planned".