Fix cursor encoding for column-form SortParam to_replace

[GayathriSrividya]

closes #67970

Summary

• update SortParam.row_value to fall back to getattr(row, name) when to_replace uses column-form mappings
• keep existing string-alias behavior unchanged
• add regression coverage in common parameter and cursor tests

Why

Task Instances sorting by run_after/logical_date/data_interval_* uses column-form to_replace mappings. Cursor encoding previously raised NotImplementedError when has_next=true, causing HTTP 500.

Validation

• /opt/homebrew/bin/ruff format + check --fix on touched files passed
• pytest execution is currently blocked in this environment by missing provider dependency: airflow.providers.openlineage (fails before tests run)

[pierrejeambrun]

Thanks for tracking this down and following the issue's analysis — the run_after 500 is real and this is close.

One thing I'd change before merge: returning getattr(row, name, None) for column-form to_replace fixes run_after/logical_date (those are association_proxy on TaskInstance), but data_interval_start/data_interval_end aren't attributes on TaskInstance, so they resolve to None. A None cursor value becomes the pagination anchor, so the next-page WHERE compares against NULL, matches nothing, and the next page comes back empty with a 200 — arguably worse than the current 500, since it silently drops rows instead of erroring. The original NotImplementedError was guarding exactly this.

Could we keep the loud failure for keys the model doesn't expose, while resolving the ones it does (see the inline suggestion)? Also looks like it needs a rebase — it's currently conflicting with main.

[potiuk]

@GayathriSrividya A few things need addressing before review — see our Pull Request quality criteria.

• ❌ Merge conflicts with main. See docs.

No rush.

Note: This comment was drafted by an AI-assisted triage tool and may contain mistakes. Once you have addressed the points above, an Apache Airflow maintainer — a real person — will take the next look at your PR. We use this two-stage triage process so that our maintainers' limited time is spent where it matters most: the conversation with you.

[GayathriSrividya]

@GayathriSrividya A few things need addressing before review — see our Pull Request quality criteria.

• ❌ Merge conflicts with main. See docs.

No rush.

Note: This comment was drafted by an AI-assisted triage tool and may contain mistakes. Once you have addressed the points above, an Apache Airflow maintainer — a real person — will take the next look at your PR. We use this two-stage triage process so that our maintainers' limited time is spent where it matters most: the conversation with you.

@potiuk Resolved — rebased on latest main, all reviewer suggestions applied.

[pierrejeambrun]

LGTM, thanks!

CI need fixing. (we'r missing an import on dag_maker)