fix schedule_downstream_tasks bug (#42582)

[potiuk]

• fix schedule_downstream_tasks bug

• remove partial_subset

• Update comment

---

Co-authored-by: 维湘 jiazhao.ljz@alibaba-inc.com
(cherry picked from commit 3fceaa6)

---

^ Add meaningful description above
Read the Pull Request Guidelines for more information.
In case of fundamental code changes, an Airflow Improvement Proposal (AIP) is needed.
In case of a new dependency, check compliance with the ASF 3rd Party License Policy.
In case of backwards incompatible changes please leave a note in a newsfragment file, named {pr_number}.significant.rst or {issue_number}.significant.rst, in newsfragments.

[potiuk]

This is backport of #42582

Since we are closing to 2.10.3 I merged #42582 as I believe it solves often occuring "serialization" issues (with partial_subset) that happen in "mini-scheduler" and causes failures of celery workers - and tests had shown that partial_subset are also wrong when it comes to determining eligibility of dependent tasks in some cases.

@uranusjr and @ashb -> if you have any concerns about it this is the time to raise it before we merge - I think you were mostly those who knew most about using partial_subset and mini-scheduler so if you have any doubts whether we should release it in 2.10.3 - please comment. Details and discussion in #42582 and in the comments in this PR.

[potiuk]

The failure here should be fixed by #43307

[potiuk]

Failing tests fixed in tip of 2-10-test merging.