[HSTACK] feat: added codec for DeltaScanMetaExec

[mihai-bleont]

This is mostly copied from the DeltaScanExec codec. The important difference is that DeltaScanMetaExec has no child plan, instead it has a pair vector of (file_id, row_count) embedded.

I tested with the following SQLs that used to fail:

SELECT _ACP_DATE, COUNT(*) FROM aep_dataset_next('4E9432245BC7C44B0A494037@AdobeOrg', 'prod', '', '68e0baa39036635052fa02cd') GROUP BY _ACP_DATE;
SELECT DISTINCT _ACP_DATE FROM aep_dataset_next('4E9432245BC7C44B0A494037@AdobeOrg', 'prod', '', '68e0baa39036635052fa02cd');
SELECT _ACP_DATE FROM aep_dataset_next('4E9432245BC7C44B0A494037@AdobeOrg', 'prod', '', '68e0baa39036635052fa02cd');
SELECT _ACP_DATE, COUNT(*) FROM aep_dataset_next('4E9432245BC7C44B0A494037@AdobeOrg', 'prod', '', '68e0baa39036635052fa02cd')
WHERE _ACP_DATE = '2025-10-04' GROUP BY _ACP_DATE;

SELECT r.timestamp, r._id, n._ACP_DATE, n.cnt
FROM (
  SELECT timestamp, _id
  FROM aep_dataset('4E9432245BC7C44B0A494037@AdobeOrg','prod','','abfss://gen1@datalakeg462p5ax3pjbzvdp.dfs.core.windows.net/platform/5fc8afc657486818dd2c03ed-v1/')
  LIMIT 1
) r
CROSS JOIN (
  SELECT _ACP_DATE, COUNT(*) AS cnt
  FROM aep_dataset_next('4E9432245BC7C44B0A494037@AdobeOrg', 'prod', '', '68e0baa39036635052fa02cd')
  GROUP BY _ACP_DATE
) n

[github-actions]

ACTION NEEDED

delta-rs follows the Conventional Commits specification for release automation.

The PR title and description are used as the merge commit message. Please update your PR title and description to match the specification.

[aditanase]

@mihai-bleont have we considered pushing this change upstream? (together with the other one? cc @adragomir )

[aditanase]

I think we'll also use "with files" boolean flag pretty soon

[mihai-bleont]

I'm a bit out of context with one. @aditanase is there anything we want to address in this PR?

[aditanase]

do we know how big this is? in general, do we have a metric for the plan size?
Ideally this would be extremely small, relying solely on metadata and with values burned in, I'd like to check though that for a really large table we are not carrying over all of the metadata in JSON form

[mihai-bleont]

This has been merged in c76cb33